WO2018196553A1 - Method and apparatus for obtaining identifier, storage medium, and electronic device - Google Patents

Abstract

Disclosed are a method and apparatus for obtaining an identifier, a storage medium, and an electronic device. The method comprises: obtaining identifiers corresponding to a predetermined operation from multiple data sources; obtaining initial identifiers from among the identifiers according to characteristic information of the identifiers and preset characteristic words; determining characteristic parameters of the initial identifiers according to a preset weight and the characteristic information; obtaining a first target identifier from among the initial identifiers, wherein the first target identifier is a collection of identifiers of which the characteristic parameters are higher than a preset parameter in the initial identifiers. By means of the technical solution, the technical problem in the related art of low accuracy for obtaining an identifier for training is resolved.

Description

Method and device for acquiring identification, storage medium and electronic device

The present application claims the priority of the Chinese Patent Application, filed on Apr. 27, 2017, the disclosure of which is incorporated herein by reference.

Technical field

The present invention relates to the field of computers, and in particular to a method and device for acquiring an identifier, a storage medium, and an electronic device.

Background technique

In many recommended areas, such as advertising recommendations, game recommendations, video recommendations, news recommendations, etc., it is often necessary to devote resources to a specific domain of users (targeted users) to improve the effectiveness of resource delivery, while targeted user mining, usually using training The way to predict the model, including Logistic Regression (LR), Random Forest (RF), Gradient Boosting Decision Tree (GBDT), etc. The key to bad is to select the accuracy of the training sample (which can be the user's logo) during its training phase, that is, whether the positive and negative sample selection is accurate enough. The method of obtaining a true positive sample is to obtain real and reliable positive sample data according to Customer Relationship Management (CRM), but often the size of such data is small, resulting in the characteristics of the trained model are not obvious enough. Thereby affecting the model training effect.

Most of the existing training data sample acquisition methods are based on user behaviors to obtain rule matching populations from a single data source. As a positive sample set, negative sample sets are randomly selected from the market; this way the single data source is easy. The sample is biased and the size of the generated sample set is relatively small. In addition, the sample set selected is not easy to distinguish the purity of each sample.

In the existing method of acquiring training data samples, as shown in FIG. 1 , according to a specific group to be mined, sample representation words and optimization rules are prepared, and in a single user behavior log, pattern matching (regular matching) is used to mine The population with the characteristics of the sample character is selected as the positive sample population of the training data, and the negative sample population is the randomly selected sample after the positive sample population is excluded from the large population. This approach will lead to the following defects: First, the user behavior log is single, the search matching population is limited, and the sample is biased. Secondly, the positive sample population is not enough to explain the purity and reliability of the positive sample after mining through pattern matching. The above defects lead to the acquisition of existing training data samples in a manner that obtains less accurate identification of the training.

In response to the above problems, no effective solution has been proposed yet.

Summary of the invention

The embodiment of the invention provides a method and a device for acquiring an identifier, a storage medium and an electronic device, so as to at least solve the technical problem that the accuracy of obtaining the identifier for training in the related art is low.

According to an aspect of the embodiments of the present invention, a method for obtaining an identifier includes: obtaining an identifier corresponding to a predetermined operation from a plurality of data sources, wherein recording is performed in a target data source included in the plurality of data sources The account corresponding to the identifier and the predetermined operation performed by the account; obtaining an initial identifier from the identifier according to the feature information of the identifier and a preset feature word, wherein the feature information is used to indicate a feature of the predetermined operation; determining a feature parameter of the initial identifier according to the preset weight and the feature information, wherein the preset weight corresponds to the target data source, and the preset weight is used to indicate the The frequency of the predetermined operation performed by the account in the target data source, the feature parameter is used to indicate the frequency at which the initial identifier performs the predetermined operation; and the first target identifier is obtained from the initial identifier, where the first A target identifier is a set of identifiers in the initial identifier that are higher than a preset parameter.

According to another aspect of the present invention, an apparatus for acquiring an identifier is provided, including: a first acquiring module, configured to acquire an identifier corresponding to a predetermined operation from a plurality of data sources, wherein The target data source included in the data source includes the account corresponding to the identifier and the predetermined operation performed by the account; the second obtaining module is configured to set the feature information according to the identifier and the preset feature word Obtaining an initial identifier from the identifier, wherein the feature information is used to represent a feature of the predetermined operation; and the determining module is configured to determine a feature parameter of the initial identifier according to the preset weight and the feature information, where The preset weight is used to indicate the frequency at which the account in the target data source performs the predetermined operation, and the feature parameter is used to indicate that the initial identifier is executed. a frequency of the predetermined operation; a third obtaining module configured to acquire a first target identifier from the initial identifier, wherein the first target identifier And is a set of identifiers in the initial identifier that are higher than a preset parameter.

According to another aspect of the embodiments of the present invention, there is also provided a storage medium, the storage medium comprising a stored program, wherein, when the program is running, controlling a device in which the storage medium is located to perform an acquisition method of the identifier.

In the embodiment of the present invention, the identifier corresponding to the predetermined operation is obtained from the plurality of data sources, wherein the target data source included in the plurality of data sources records the predetermined operation performed by the account and the account corresponding to the identifier; The identifier information and the preset feature word obtain an initial identifier from the identifier, wherein the feature information is used to represent a feature of the predetermined operation; and the feature parameter of the initial identifier is determined according to the preset weight and the feature information, wherein the preset weight and the target data are Correspondingly, the preset weight is used to indicate the frequency at which the account in the target data source performs the predetermined operation, the feature parameter is used to indicate the frequency at which the initial identifier performs the predetermined operation, and the first target identifier is obtained from the initial identifier, where the first target identifier It is a set of identifiers in which the feature parameter is higher than the preset parameter in the initial identifier. That is to say, in the target data source, the account corresponding to the identifier and the predetermined operation performed by the account are recorded, and the identifier corresponding to the predetermined operation is obtained, so that the obtaining path of the identifier is more extensive, and the logo size is avoided from a single user log. If the identifier of the acquired identifier is biased, the initial identifier is initially filtered according to the feature information of the identifier and the preset feature word, and the feature identifier is determined according to the preset weight and the feature information to indicate the initial identifier execution. The frequency of the predetermined operation is then obtained from the initial identifier, the first target flag whose feature parameter is higher than the preset parameter, so that the identifier included in the first target identifier is an identifier that performs a predetermined operation frequency, thereby improving the acquisition. The accuracy of the identification of the training further overcomes the problem of low accuracy in obtaining the identification for training in the related art.

DRAWINGS

The drawings described herein are intended to provide a further understanding of the invention, and are intended to be a part of the invention. In the drawing:

1 is a schematic diagram of an acquisition method of an identifier according to the related art;

2 is a schematic diagram of an application environment of an optional method for acquiring an identifier according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of an optional method for acquiring an identifier according to an embodiment of the present invention; FIG.

4 is a schematic diagram 1 of an optional identification acquiring device according to an embodiment of the present invention;

FIG. 5 is a second schematic diagram of an apparatus for acquiring an identifier according to an embodiment of the present invention; FIG.

FIG. 6 is a third schematic diagram of an optional identification acquiring device according to an embodiment of the present invention; FIG.

FIG. 7 is a schematic diagram 4 of an apparatus for acquiring an optional identifier according to an embodiment of the present invention; FIG.

FIG. 8 is a schematic diagram 5 of an optional identifier acquiring apparatus according to an embodiment of the present invention; FIG.

9 is a schematic diagram 6 of an optional identification acquiring device according to an embodiment of the present invention;

FIG. 10 is a schematic diagram of an application scenario of an optional method for acquiring an identifier according to an embodiment of the present invention;

FIG. 11 is a schematic structural diagram of an electronic device according to an embodiment of the invention.

detailed description

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is an embodiment of the invention, but not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of the present invention.

It is to be understood that the terms "first", "second" and the like in the specification and claims of the present invention are used to distinguish similar objects, and are not necessarily used to describe a particular order or order. It is to be understood that the data so used may be interchanged where appropriate, so that the embodiments of the invention described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "comprises" and "comprises" and "the" and "the" are intended to cover a non-exclusive inclusion, for example, a process, method, system, product, or device that comprises a series of steps or units is not necessarily limited to Those steps or units may include other steps or units not explicitly listed or inherent to such processes, methods, products or devices.

In the embodiment of the present invention, an embodiment of a method for acquiring the foregoing identifier is provided. As an optional implementation manner, the method for obtaining the identifier may be, but is not limited to, being applied to an application environment as shown in FIG. 2, and the server 202 is configured to obtain an identifier corresponding to the predetermined operation from the plurality of data sources. Acquiring an initial identifier from the identifier according to the feature information of the identifier and the preset feature word, obtaining an initial identifier from the identifier according to the feature information of the identifier and the preset feature word, and acquiring the first target identifier from the initial identifier; wherein, the plurality of data The target data source included in the source records an operation performed by the account and the account corresponding to the identifier; the feature information is used to indicate the feature of the predetermined operation; the preset weight corresponds to the target data source, and the preset weight is used to indicate the target data source. The frequency of the predetermined operation performed by the account number, the characteristic parameter is used to indicate the frequency at which the initial identifier performs the predetermined operation; and the first target identifier is a set of the identifiers in the initial identifier whose feature parameters are higher than the preset parameters.

In this embodiment, the account corresponding to the identifier and the operation performed by the account are recorded in the target data source, and the server 202 obtains the identifier corresponding to the predetermined operation, so that the acquisition path of the identifier is more extensive, and the single user log is avoided. Obtaining the problem that the acquired identifier is biased due to the small size of the identifier, and then initially screening the initial identifier according to the feature information of the identifier and the preset feature word, and determining the feature parameter according to the preset weight and the feature information to represent the feature identifier Initially identifying the frequency of performing the predetermined operation, and then acquiring, from the initial identifier, the first target flag whose feature parameter is higher than the preset parameter, so that the identifier included in the first target identifier is an identifier that performs a predetermined operation frequency, thereby improving The accuracy of the identification for training is obtained, thereby overcoming the problem of low accuracy in obtaining the identification for training in the related art.

Optionally, in this embodiment, the server 202 is configured to: acquire the first feature word and the second feature word, where the preset feature word includes the first feature word and the second feature word; and obtain the initial identifier from the identifier The feature information corresponding to the initial identifier carries the first feature word and does not carry the second feature word.

Optionally, in this embodiment, the server 202 is configured to: obtain a preset weight, wherein a larger value of the preset weight indicates that the frequency of the account in the target data source performs a predetermined operation is higher; and the feature information is obtained from the feature information. The time information and the frequency information, wherein the time information is used to indicate the time when the performing the predetermined operation is performed, the frequency information is used to indicate the frequency of the identification performing the predetermined operation, and the characteristic parameter is determined according to the preset weight, the time information and the frequency information, wherein the characteristic parameter A larger value indicates that the initial identification performs a predetermined operation more frequently.

Optionally, in this embodiment, the server 202 is configured to: acquire a proportion of an account that performs a predetermined operation in the target data source, and allocate a preset to the target data source according to the ratio; a weight, wherein the larger the proportion, the greater the preset weight of the data source allocation; or the number of the same identifier in the first identifier set and the preset identifier set, wherein the first identifier set is a target data in the initial identifier A set of identifiers included in the source; a preset weight is assigned to the target data source according to a ratio between the quantity and the number of identifiers in the first identifier set, wherein the larger the ratio, the greater the preset weight of the data source allocation.

Optionally, in this embodiment, the server 202 is configured to: calculate a product of the corresponding time information and frequency information of the initial identifier in each target data source; calculate a weighted sum of the products according to the preset weight to obtain a feature parameter.

Optionally, in this embodiment, the server 202 is configured to: acquire information for indicating a feature of the predetermined operation from the predetermined operation corresponding to the identifier, wherein the information for indicating the feature of the predetermined operation includes: the predetermined operation corresponding to Characteristic words, time information and frequency information; storing feature words, time information and frequency information into a preset format to obtain feature information.

Optionally, in this embodiment, the server 202 is configured to: arrange the initial identifiers according to the feature parameters from high to low; select the first target identifier from the aligned identifiers, where the first target identifier is included in The identifier of the first N bits in the aligned identifiers; or the first target identifier whose value of the feature parameter is greater than or equal to the preset value is obtained from the initial identifier.

Optionally, in this embodiment, the server 202 is configured to: match the first target identifier with the preset target identifier; and determine the first target if the first target identifier and the preset target identifier match successfully. The identifier is the required identifier; if the first target identifier and the preset target identifier are unsuccessful, the first target identifier is re-acquired.

Optionally, in this embodiment, the server 202 is further configured to: determine whether the first target identifier and the preset target identifier include the same identifier that is greater than or equal to the preset number; and determine the first target identifier and the preset If the target identifier includes the same identifier that is greater than or equal to the preset number, the first target identifier is determined to be successfully matched with the preset target identifier.

Optionally, in this embodiment, the server 202 is further configured to: obtain an identifier corresponding to the account that is included in the multiple data sources; and randomly obtain the first target identifier from the identifier corresponding to the account that is included in the multiple data sources. The identifier other than the identifier is obtained, wherein the number of the identifiers included in the second target identifier is the same as the number of the identifiers included in the first target identifier.

Optionally, in the application environment described in this embodiment, the client may further include a client connected to the server 202 through a network, where the server 202 is further configured to: train the prediction model according to the first target identifier and the second target identifier; According to the prediction model, the to-be-pushed identifier is obtained for the to-be-pushed resource from the identifiers of the plurality of data sources, and the to-be-pushed resource is pushed to the client used by the account corresponding to the to-be-advertised identifier.

Optionally, in this embodiment, the foregoing client may include, but is not limited to, at least one of the following: a mobile phone, a tablet computer, a notebook computer, a desktop PC, a digital television, and other hardware devices for area sharing. The above network may include, but is not limited to, at least one of the following: a wide area network, a metropolitan area network, and a local area network. The above is only an example, and the embodiment does not limit this.

According to an embodiment of the present invention, a method for obtaining an identifier is provided. As shown in FIG. 3, the method includes:

S302, the identifier corresponding to the predetermined operation is obtained from the plurality of data sources, where the account and the account corresponding to the identifier are recorded in the target data source included in the plurality of data sources;

S304. Acquire an initial identifier from the identifier according to the identifier information and the preset feature word, where the feature information is used to indicate a feature of the predetermined operation.

S306. Determine a feature parameter of the initial identifier according to the preset weight and the feature information, where the preset weight is corresponding to the target data source, and the preset weight is used to indicate a frequency at which the account in the target data source performs a predetermined operation, and the feature parameter is used to indicate Initially identifying the frequency at which the predetermined operation is performed;

S308. Acquire a first target identifier from the initial identifier, where the first target identifier is a set of identifiers in the initial identifier that have a feature parameter higher than a preset parameter.

Optionally, in this embodiment, the method for acquiring the identifier may be, but is not limited to, being applied to the method of acquiring the identifier sample for model training, and using the training result to push the resource for the client. The above client may be, but not limited to, various types of software, such as search software, social software, instant messaging software, news information software, game software, shopping software, and the like. Optionally, it may be, but is not limited to, being applied to a scenario in which the foregoing identification sample is used for model training, and the training result is used by the client of the shopping software to push resources, or may be, but not limited to, applied to the model training in the above-mentioned acquisition identification sample. The training result is used to push the resource of the client of the search software to realize the acquisition of the identification sample. The above is only an example, and is not limited in this embodiment.

Optionally, in this embodiment, the multiple data sources may be various platforms, software, websites, applications, and the like. For example: social applications, search engines, e-commerce websites, advertising platforms, etc.

Optionally, in this embodiment, the identifiers may correspond to different account accounts in different data sources. For example, a user may have registered an account on multiple applications, for example, an account A is registered on the social platform, an account B is registered on the shopping website, and an account C is registered on the instant messaging application, and the user can If the three accounts on the above platform are associated, the three accounts A, B, and C can be used to uniquely identify the user.

Optionally, in this embodiment, one or more data sources may be included in the target data source. That is to say, the account in the data source corresponding to the identifier is recorded in the data source, and the operation performed by the account. The identifier corresponding to the predetermined operation may be recorded in one of the plurality of data sources, or may be recorded in several of the plurality of data sources.

Optionally, in this embodiment, the predetermined operation may be to identify a certain behavior performed or a phrase for characterizing the behavior. For example, if the user to be excavated is a user who purchases a maternal and child product, the predetermined operation may be "click on the entry with milk powder or diaper", or "milk powder", "diaper" and the like. The identifier corresponding to the predetermined operation obtained from the plurality of data sources may first obtain an account searched for “milk powder” and “diaper” in the search engine, and the account of the purchased milk powder or diaper in the shopping website is sent in the instant messaging software. The account number of the message "milk powder", "diaper" and the like, and the account number of the item with the powdered milk or the diaper are clicked in the multiple data sources, and the corresponding identifiers of the above-mentioned accounts are obtained.

Optionally, in this embodiment, the initial identifier may include, but is not limited to, including one or more identifiers. The preset feature words may be, but are not limited to, one or more feature words. The first target identifier may include, but is not limited to, including one or more identifiers.

Optionally, in this embodiment, the preset weight may be used to indicate a frequency at which an account in the target data source performs a predetermined operation. In other words, the preset weight can be used to indicate the degree of attention of the account in the target data source to the predetermined operation, which can be represented by, but not limited to, the frequency with which the account in the target data source performs the predetermined operation. Here, the frequency at which the account in the target data source performs the predetermined operation may be, but is not limited to, the number of accounts in the target data source that are frequently performed by the predetermined operation (for example, the account whose frequency exceeds 5 times per day performs the predetermined operation. 50% of the total number of accounts in the data source). Alternatively, but not limited to, performing the saliency of the predetermined operation with the account number in the target data source to indicate the frequency at which the account in the target data source performs the predetermined operation. The significance of the account performing the predetermined operation in the target data source can be determined by calculating the proportion of the identifier in the initial identifier in which the account number is recorded in the target data source (for example, the identifier of the last push resource).

Optionally, in this embodiment, the preset weight may be set according to a frequency at which the predetermined operation is performed by the account in the target data source, or may be performed according to the frequency of performing the predetermined operation according to the account in the target data source. The model is trained in the way it is calculated.

It can be seen that, through the foregoing steps, the account corresponding to the identifier and the operation performed by the account are recorded in the target data source, and the identifier corresponding to the predetermined operation is obtained from the target data source, so that the obtaining path of the identifier is more extensive, and the identifier is obtained from a single user log. The acquired identifier has a biased problem, and the initial identifier is preliminarily selected according to the feature information of the identifier and the preset feature word, and the feature identifier is determined according to the preset weight and the feature information to represent the initial identifier. The frequency of the predetermined operation is performed, and then the first target flag whose feature parameter is higher than the preset parameter is obtained from the initial identifier, so that the identifier included in the first target identifier is an identifier that performs a predetermined operation frequency, thereby improving acquisition. The accuracy of the identification for training further overcomes the problem of low accuracy in obtaining the identification for training in the related art.

As an optional solution, obtaining the initial identifier from the identifier according to the identifier information and the preset feature word includes:

S1: acquiring a first feature word and a second feature word, where the preset feature word includes a first feature word and a second feature word;

S2. The initial identifier is obtained from the identifier, where the feature information corresponding to the initial identifier carries the first feature word and does not carry the second feature word.

Optionally, in this embodiment, the preset feature words may include, but are not limited to, a first feature word and a second feature word. The preset feature words can be used to represent characteristics of a type of user population, which can include positive representation words and negative representation words, wherein the positive representation words (equivalent to the first characteristic words described above), that is, the key words in the popular sense Words, used to represent feature populations, negative representation words (equivalent to the above second feature words), ie filter words (filter_words), the role of negative representation words, in denoising, that is, to remove some multi-word stitching The latter noise makes the positive representation words more representative of the characteristic population.

Through the above steps, the initial identification of the identifier is achieved by obtaining the initial identifier from the identifier according to the identified feature information and the first feature word and the second feature word included in the preset feature word.

As an optional solution, determining the feature parameters of the initial identifier according to the preset weight and the feature information includes:

S1: Obtain a preset weight, wherein a larger value of the preset weight indicates that the frequency of performing the predetermined operation on the account in the target data source is higher;

S2: Obtain time information and frequency information from the feature information, where the time information is used to indicate a time when the performing a predetermined operation is performed, and the frequency information is used to indicate a frequency indicating that the predetermined operation is performed;

S3. The feature parameter is determined according to the preset weight, the time information, and the frequency information. The greater the value of the feature parameter, the higher the frequency at which the initial identifier performs the predetermined operation.

Optionally, in this embodiment, the preset weight may be obtained by one of the following methods:

In the first manner, the proportion of the account that performs the predetermined operation in the target data source in all the accounts included in the target data source is obtained; the preset weight is assigned to the target data source according to the ratio, wherein the larger the proportion of the data source is allocated Set the weight more.

For example, there are three target data sources: target data source A, target data source B, and target data source C. There are 100 accounts in the target data source A, and 34 of them have performed predetermined operations on the target data source. There are 200 accounts in B, of which 25 have performed predetermined operations, and there are 100 accounts in the target data source C, of which 56 have performed predetermined operations. Then, the ratios of the target data source A, the target data source B, and the target data source C are 34%, 12.5%, and 56%, respectively, and the target data source A, the target data source B, and the target data are obtained according to the obtained ratio. Source C assigns preset weights 2, 1, and 3.

Manner 2: Obtain a quantity of the same identifier in the first identifier set and the preset identifier set, where the first identifier set is a set of identifiers included in a target data source in the initial identifier; and the identifier in the first identifier set according to the quantity The ratio between the number of inputs is a preset weight assigned to the target data source, wherein the larger the ratio, the greater the default weight assigned by the data source.

Optionally, in this embodiment, the preset identifier set may be, but is not limited to, an identifier included in the target data source in the first target identifier acquired in the previous time, or is included in the target data source according to the identifier of the previous push data. Logo.

In an optional implementation manner, the identifier of the target data source in the first target identifier acquired by the preset identifier set is taken as an example, and the preset identifier set A corresponding to the target data source A includes 40 identifiers, and the target data is included. The preset identifier set B corresponding to the source B includes 30 identifiers, and the preset identifier set C corresponding to the target data source C includes 40 identifiers; the initial identifier includes the target data source A, the target data source B, and the target data. The number of the identifiers of the source C is 20, 40, and 40 respectively. Then, the first identifier set A corresponding to the target data source A includes 20 identifiers, and the first identifier set B corresponding to the target data source B includes 40 identifiers. The first identifier set C corresponding to the target data source C includes 40 identifiers, wherein the first identifier set A is matched with the identifier in the preset identifier set A, and the first identifier set A and the preset identifier set A are obtained. The number of the same identifier is 10, and the first identifier set B is matched with the identifier in the preset identifier set B, and the number of the same identifier in the first identifier set B and the preset identifier set B is 5, which will be the first Standard The set C is matched with the identifier in the preset identifier set C, and the number of the same identifiers in the first identifier set C and the preset identifier set C is 20, and the target data source A is obtained according to the obtained number of the same identifiers. The target data source B and the target data source C are respectively assigned preset weights 2, 1, and 3.

Optionally, in this embodiment, the feature parameter may be determined by: calculating a product of the corresponding time information and frequency information of the initial identifier in each target data source, and then calculating a weighted sum of the products according to the preset weight, to obtain Characteristic Parameters.

In an alternative embodiment, the above characteristic parameters can be calculated by the following formula:

Among them, source represents the data source, there are n data sources; weight represents the preset weight on each data source; time represents the above time information, you can use abs (user behavior occurrence time - current mining time ), that is, the absolute value of the behavior time difference to represent the above time information, as the user behavior time decay parameter, that is, the closer the behavior occurs to the current time, the larger the feature parameter is, the farther from the current time, the smaller the feature parameter; Representing the above frequency information, it can be used to indicate the frequency of user behavior. Here, the sigmoid function is taken and normalized; the more the behavior frequency is, the higher the feature parameter is.

It can be seen that, by using the foregoing steps, the feature identifier of the initial identifier is determined according to the preset weight and the feature information, and the initial identifier is scored, which can be used to measure the frequency at which the initial identifier performs a predetermined operation, so that the first target identifier selected from the initial identifier is further It can represent a predetermined operation, thereby improving the accuracy of obtaining the identification for training, thereby overcoming the problem of low accuracy in obtaining the identification for training in the related art.

As an optional solution, before the initial identifier is obtained from the identifier according to the identifier information and the preset feature word, the method further includes:

S1. Acquire information for indicating a feature of the predetermined operation from a predetermined operation corresponding to the identifier, where the information for indicating the feature of the predetermined operation includes: a feature word corresponding to the predetermined operation, time information, and frequency information;

S2: storing feature words, time information, and frequency information into a preset format to obtain feature information.

It can be seen that, through the above steps, the information for indicating the feature of the predetermined operation acquired from the predetermined operation corresponding to the identification is organized into a predetermined format for storage, so that the comparison of the feature words is faster and more convenient.

As an optional solution, obtaining the first target identifier from the initial identifier includes one of the following:

S1, the initial identifiers are arranged according to the feature parameters from high to low; the first target identifier is selected from the aligned identifiers, wherein the first target identifier includes the identifiers ranked in the first N digits in the aligned identifiers;

S2. Acquire, from the initial identifier, a first target identifier whose value of the feature parameter is greater than or equal to a preset value.

Optionally, in this embodiment, the feature parameters may be sorted from high to low, and the identifiers ranked in the first N bits are used as identifiers whose feature parameters are higher than the preset parameters, to obtain the first target identifier.

Optionally, in this embodiment, the preset value may be set, and the identifier corresponding to the feature parameter whose value is greater than or equal to the preset value is used as the first target identifier.

It can be seen that, through the above steps, obtaining the first target identifier by sorting the feature parameters from high to low, or setting the preset value, can clearly select an identifier that is more representative of the predetermined operation from the initial identifier.

As an optional solution, after obtaining the first target identifier from the initial identifier, the method further includes:

S1, matching the first target identifier with the preset target identifier;

S2. If the first target identifier and the preset target identifier are successfully matched, determine that the first target identifier is the required identifier; if the first target identifier and the preset target identifier are unsuccessful, re-acquire the first A target identifier.

Optionally, in this embodiment, the first target identifier is matched with the preset target identifier by determining whether the first target identifier and the preset target identifier include the same identifier that is greater than or equal to the preset number. If it is determined that the first target identifier and the preset target identifier include the same identifier that is greater than or equal to the preset number, the first target identifier is determined to be successfully matched with the preset target identifier.

Optionally, in this embodiment, the preset target identifier may be the first target identifier acquired last time, and may also be a preset target identifier.

Optionally, in this embodiment, when the first target identifier is re-acquired, the first target identifier may be obtained by, but not limited to, re-acquiring the identifier corresponding to the predetermined operation by resetting the predetermined operation. It is also possible, but not limited to, to reacquire the first target identity by re-assigning the target weight to the target data source.

It can be seen that, by using the foregoing steps, the first target identifier is matched with the preset target identifier. If the matching succeeds, it may be determined that the currently acquired first target identifier meets the needs of the model training, that is, the first target identifier is required. Logo. On the other hand, if the matching is unsuccessful, it indicates that the currently acquired first target identifier does not meet the needs of the model training, and the first target identifier may be reacquired.

As an optional solution, after obtaining the first target identifier from the initial identifier, the method further includes:

S1. Acquire an identifier corresponding to an account included in multiple data sources.

S2, randomly obtaining an identifier other than the first target identifier from the identifiers corresponding to the account numbers included in the multiple data sources, to obtain a second target identifier, where the number of the identifiers included in the second target identifier is the first target identifier The number of tags included in the same is the same.

Optionally, in this embodiment, the first target identifier may be used as a positive sample of the model training, and after acquiring the first target identifier, the second target identifier may be obtained as a model from all identifiers of the multiple data sources. Negative sample of training.

As an optional solution, the identifiers other than the first target identifier are randomly obtained from the identifiers corresponding to the account numbers included in the multiple data sources, and after the second target identifier is obtained, the method further includes:

S1, training a prediction model according to the first target identifier and the second target identifier;

S2. Acquire, according to the prediction model, the to-be-pushed identifier for the to-be-pushed resource from the identifiers included in the multiple data sources;

S3: Push the to-be-pushed resource to the to-be-pushed identifier.

Optionally, in this embodiment, the acquired first target identifier and the second target identifier may be used to perform training of the predictive model, so that the to-be-pushed identifier obtained by the predictive model can more accurately represent the pointed operation. crowd. Thereby, the efficiency of pushing resources can be made higher.

It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations, but those skilled in the art should understand that the present invention is not limited by the described action sequence. Because certain steps may be performed in other sequences or concurrently in accordance with the present invention. In addition, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

Through the description of the above embodiments, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, by hardware, but in many cases, the former is A better implementation. Based on such understanding, the technical solution of the present invention in essence or the contribution to the related art can be embodied in the form of a software product stored in a storage medium (such as ROM/RAM, disk, CD-ROM). The instructions include a number of instructions for causing a terminal device (which may be a cell phone, computer, server, or network device, etc.) to perform the methods described in various embodiments of the present invention.

According to an embodiment of the present invention, there is also provided an apparatus for acquiring an identifier of an acquisition method of the foregoing identifier. As shown in FIG. 4, the apparatus includes:

1) The first obtaining module 42 is configured to obtain an identifier corresponding to the predetermined operation from the plurality of data sources, wherein the account and the account corresponding to the identifier are recorded in the target data source included in the plurality of data sources operating;

2) The second obtaining module 44 is configured to obtain an initial identifier from the identifier according to the identified feature information and the preset feature word, where the feature information is used to indicate a feature of the predetermined operation;

The determining module 46 is configured to determine the feature parameter of the initial identifier according to the preset weight and the feature information, wherein the preset weight corresponds to the target data source, and the preset weight is used to indicate that the account in the target data source performs the predetermined operation. Frequency, the characteristic parameter is used to indicate the frequency at which the initial identification performs a predetermined operation;

4) The third obtaining module 48 is configured to obtain the first target identifier from the initial identifier, where the first target identifier is a set of identifiers in the initial identifier whose feature parameters are higher than the preset parameters.

Optionally, in this embodiment, the acquiring device of the foregoing identifier may be, but is not limited to, being applied to the method of acquiring the identifier sample for model training, and using the training result to push the resource for the client. The above client may be, but not limited to, various types of software, such as search software, social software, instant messaging software, news information software, game software, shopping software, and the like. Optionally, it may be, but is not limited to, being applied to a scenario in which the foregoing identification sample is used for model training, and the training result is used by the client of the shopping software to push resources, or may be, but not limited to, applied to the model training in the above-mentioned acquisition identification sample. The training result is used to push the resource of the client of the search software to realize the acquisition of the identification sample. The above is only an example, and is not limited in this embodiment.

Optionally, in this embodiment, the multiple data sources may be various platforms, software, websites, applications, and the like. For example: social applications, search engines, e-commerce websites, advertising platforms, etc.

Optionally, in this embodiment, the identifiers may correspond to different account accounts in different data sources. For example, a user may have registered an account on multiple applications, for example, an account A is registered on the social platform, an account B is registered on the shopping website, and an account C is registered on the instant messaging application, and the user can If the three accounts on the above platform are associated, the three accounts A, B, and C can be used to uniquely identify the user.

Optionally, in this embodiment, one or more data sources may be included in the target data source. That is to say, the account in the data source corresponding to the identifier is recorded in the data source, and the operation performed by the account. The identifier corresponding to the predetermined operation may be recorded in one of the plurality of data sources, and may also be recorded in several of the plurality of data sources.

Optionally, in this embodiment, the predetermined operation may be to identify a certain behavior performed or a phrase for characterizing the behavior. For example, if the user to be excavated is a user who purchases a maternal and child product, the predetermined operation may be "click on the entry with milk powder or diaper", or "milk powder", "diaper" and the like. The identifier corresponding to the predetermined operation obtained from the plurality of data sources may first obtain an account searched for “milk powder” and “diaper” in the search engine, and the account of the purchased milk powder or diaper in the shopping website is sent in the instant messaging software. The account number of the message "milk powder", "diaper" and the like, and the account number of the item with the powdered milk or the diaper are clicked in the multiple data sources, and the corresponding identifiers of the above-mentioned accounts are obtained.

Optionally, in this embodiment, the initial identifier may include, but is not limited to, including one or more identifiers. The preset feature words may be, but are not limited to, one or more feature words. The first target identifier may include, but is not limited to, including one or more identifiers.

Optionally, in this embodiment, the preset weight may be used to indicate a frequency at which an account in the target data source performs a predetermined operation. In other words, the preset weight can be used to indicate the degree of attention of the account in the target data source to the predetermined operation, which can be expressed, but not limited to, by the frequency with which the account in the target data source performs the predetermined operation. Here, the frequency at which the account in the target data source performs the predetermined operation may be, but is not limited to, the number of accounts in the target data source that are frequently performed by the predetermined operation (for example, the account whose frequency exceeds 5 times per day performs the predetermined operation. 50% of the total number of accounts in the data source). Alternatively, but not limited to, performing the saliency of the predetermined operation with the account number in the target data source to indicate the frequency at which the account in the target data source performs the predetermined operation. The significance of the account performing the predetermined operation in the target data source can be determined by calculating the proportion of the identifier in the initial identifier in which the account number is recorded in the target data source (for example, the identifier of the last push resource).

Optionally, in this embodiment, the preset weight may be set according to a frequency at which the predetermined operation is performed by the account in the target data source, or may be performed according to the frequency of performing the predetermined operation according to the account in the target data source. The model is trained in the way it is calculated.

It can be seen that, by using the foregoing device, the account corresponding to the identifier and the operation performed by the account are recorded in the target data source, and the identifier corresponding to the predetermined operation is obtained, so that the obtaining path of the identifier is more extensive, and the identifier is obtained from a single user log. The acquired identifier has a biased problem, and the initial identifier is preliminarily selected according to the feature information of the identifier and the preset feature word, and the feature identifier is determined according to the preset weight and the feature information to represent the initial identifier. The frequency of the predetermined operation is performed, and then the first target flag whose feature parameter is higher than the preset parameter is obtained from the initial identifier, so that the identifier included in the first target identifier is an identifier that performs a predetermined operation frequency, thereby improving acquisition. The accuracy of the identification for training further overcomes the problem of low accuracy in obtaining the identification for training in the related art.

As an alternative, as shown in FIG. 5, the second obtaining module 44 includes:

1) The first obtaining unit 52 is configured to acquire the first feature word and the second feature word, wherein the preset feature word includes the first feature word and the second feature word;

2) The second obtaining unit 54 is configured to obtain an initial identifier from the identifier, where the feature information corresponding to the initial identifier carries the first feature word and does not carry the second feature word.

Optionally, in this embodiment, the preset feature words may include, but are not limited to, a first feature word and a second feature word. The preset feature words can be used to represent characteristics of a type of user population, which can include positive representation words and negative representation words, wherein the positive representation words (equivalent to the first characteristic words described above), that is, the key words in the popular sense Words, used to represent feature populations, negative representation words (equivalent to the above second feature words), ie filter words (filter_words), the role of negative representation words, in denoising, that is, to remove some multi-word stitching The latter noise makes the positive representation words more representative of the characteristic population.

Through the foregoing device, the initial identifier is obtained from the identifier according to the feature information of the identifier and the first feature word and the second feature word included in the preset feature word, thereby implementing preliminary screening of the identifier.

As an alternative, as shown in FIG. 6, the determining module 46 includes:

1) The third obtaining unit 62 is configured to acquire a preset weight, wherein a larger value of the preset weight indicates that the frequency of performing the predetermined operation by the account in the target data source is higher;

2) The fourth obtaining unit 64 is configured to obtain the time information and the frequency information from the feature information, wherein the time information is used to indicate the time when the performing the predetermined operation is performed, and the frequency information is used to indicate the frequency of the identification performing the predetermined operation;

3) The determining unit 66 is configured to determine the feature parameter according to the preset weight, the time information, and the frequency information, wherein the larger the value of the feature parameter is, the higher the frequency at which the initial identifier performs the predetermined operation.

Optionally, in this embodiment, the third obtaining unit 62 is set to one of the following:

Obtaining a proportion of all accounts included in the target data source in the target data source; assigning a preset weight to the target data source according to the ratio, wherein the larger the proportion of the data source, the more the preset weight is assigned Big;

Obtaining, by the first identifier set, a quantity of the same identifier in the preset identifier set, where the first identifier set is a set of identifiers included in a target data source in the initial identifier; and according to the quantity and the quantity identified in the first identifier set The ratio between the two is assigned a preset weight, and the larger the ratio, the larger the default weight assigned by the data source.

For example, there are three target data sources: target data source A, target data source B, and target data source C. There are 100 accounts in the target data source A, and 34 of them have performed predetermined operations on the target data source. There are 200 accounts in B, of which 25 have performed predetermined operations, and there are 100 accounts in the target data source C, of which 56 have performed predetermined operations. Then, the ratios of the target data source A, the target data source B, and the target data source C are 34%, 12.5%, and 56%, respectively, and the target data source A, the target data source B, and the target data are obtained according to the obtained ratio. Source C assigns preset weights 2, 1, and 3.

Optionally, in this embodiment, the preset identifier set may be, but is not limited to, an identifier included in the target data source in the first target identifier acquired in the previous time, or is included in the target data source according to the identifier of the previous push data. Logo.

In an optional implementation manner, the identifier of the target data source in the first target identifier acquired by the preset identifier set is taken as an example, and the preset identifier set A corresponding to the target data source A includes 40 identifiers, and the target data is included. The preset identifier set B corresponding to the source B includes 30 identifiers, and the preset identifier set C corresponding to the target data source C includes 40 identifiers; the initial identifier includes the target data source A, the target data source B, and the target data. The number of the identifiers of the source C is 20, 40, and 40 respectively. Then, the first identifier set A corresponding to the target data source A includes 20 identifiers, and the first identifier set B corresponding to the target data source B includes 40 identifiers. The first identifier set C corresponding to the target data source C includes 40 identifiers, wherein the first identifier set A is matched with the identifier in the preset identifier set A, and the first identifier set A and the preset identifier set A are obtained. The number of the same identifier is 10, and the first identifier set B is matched with the identifier in the preset identifier set B, and the number of the same identifier in the first identifier set B and the preset identifier set B is 5, which will be the first Standard The set C is matched with the identifier in the preset identifier set C, and the number of the same identifiers in the first identifier set C and the preset identifier set C is 20, and the target data source A is obtained according to the obtained number of the same identifiers. The target data source B and the target data source C are respectively assigned preset weights 2, 1, and 3.

Optionally, in this embodiment, the fourth obtaining unit 64 is configured to: calculate a product of the initial information and the frequency information corresponding to the initial identifier in each target data source; calculate a weighted sum of the products according to the preset weight, to obtain Characteristic Parameters.

In an alternative embodiment, the above characteristic parameters can be calculated by the following formula:

Among them, source represents the data source, there are n data sources; weight represents the preset weight on each data source; time represents the above time information, you can use abs (user behavior occurrence time - current mining time ), that is, the absolute value of the behavior time difference to represent the above time information, as the user behavior time decay parameter, that is, the closer the behavior occurs to the current time, the larger the feature parameter is, the farther from the current time, the smaller the feature parameter; Representing the above frequency information, it can be used to indicate the frequency of user behavior. Here, the sigmoid function is taken and normalized; the more the behavior frequency is, the higher the feature parameter is.

It can be seen that, by using the foregoing apparatus, the feature identifier of the initial identifier is determined according to the preset weight and the feature information, and the initial identifier is scored, which can be used to measure the frequency at which the initial identifier performs the predetermined operation, so that the first target identifier selected from the initial identifier is further It can represent a predetermined operation, thereby improving the accuracy of obtaining the identification for training, thereby overcoming the problem of low accuracy in obtaining the identification for training in the related art.

Optionally, in this embodiment, the apparatus further includes:

a sixth obtaining module, configured to acquire information for indicating a feature of the predetermined operation from a predetermined operation corresponding to the identifier, wherein the information for indicating the feature of the predetermined operation comprises: a feature word corresponding to the predetermined operation, time information, and frequency information;

The storage module is configured to store the feature words, the time information, and the frequency information into a preset format to obtain the feature information.

It can be seen that, by the above-mentioned means, the information for indicating the feature of the predetermined operation acquired from the predetermined operation corresponding to the identification is sorted into a predetermined format for storage, thereby making the comparison of the feature words more convenient and convenient.

As an alternative, as shown in FIG. 7, the third obtaining module 48 includes one of the following:

1) The processing unit 72 is configured to arrange the initial identifiers according to the feature parameters from high to low; select the first target identifier from the arranged identifiers, wherein the first target identifier is included in the aligned identifiers. The first N digits of the logo;

2) The fifth obtaining unit 74 is configured to acquire, from the initial identifier, a first target identifier whose value of the feature parameter is greater than or equal to a preset value.

Optionally, in this embodiment, the feature parameters may be sorted from high to low, and the identifiers ranked in the first N bits are used as identifiers whose feature parameters are higher than the preset parameters, to obtain the first target identifier.

Optionally, in this embodiment, the preset value may be set, and the identifier corresponding to the feature parameter whose value is greater than or equal to the preset value is used as the first target identifier.

It can be seen that, by using the above device, obtaining the first target identifier by sorting the feature parameters from high to low, or setting the preset value, it is possible to clearly select an identifier that is more representative of the predetermined operation from the initial identifier.

As an alternative, as shown in FIG. 8, the foregoing apparatus further includes:

1) The matching module 82 is configured to match the first target identifier with the preset target identifier;

2) The processing module 84 is configured to determine that the first target identifier is a required identifier if the first target identifier and the preset target identifier are successfully matched; and the first target identifier and the preset target identifier are not successfully matched. In the case of re-acquiring the first target identifier.

Optionally, in the embodiment, the matching module 82 is configured to: determine whether the first target identifier and the preset target identifier include the same identifier that is greater than or equal to the preset number; and determine the first target identifier and the preset If the target identifier includes the same identifier that is greater than or equal to the preset number, the first target identifier is determined to be successfully matched with the preset target identifier.

Optionally, in this embodiment, the preset target identifier may be the first target identifier acquired last time, and may also be a preset target identifier.

Optionally, in this embodiment, when the first target identifier is re-acquired, the first target identifier may be obtained by, but not limited to, re-acquiring the identifier corresponding to the predetermined operation by resetting the predetermined operation. It is also possible, but not limited to, to reacquire the first target identity by re-assigning the target weight to the target data source.

It can be seen that, by using the foregoing device, the first target identifier is matched with the preset target identifier. If the matching succeeds, it may be determined that the currently acquired first target identifier meets the needs of the model training, that is, the first target identifier is required. Logo. On the other hand, if the matching is unsuccessful, it indicates that the currently acquired first target identifier does not meet the needs of the model training, and the first target identifier may be reacquired.

As an alternative, as shown in FIG. 9, the foregoing apparatus further includes:

1) The fourth obtaining module 92 is configured to acquire an identifier corresponding to the account included in the plurality of data sources;

2) The fifth obtaining module 94 is configured to randomly obtain an identifier other than the first target identifier from the identifier corresponding to the account number included in the plurality of data sources, to obtain a second target identifier, where the second target identifier includes The number of identifiers is the same as the number of identifiers included in the first target identifier.

Optionally, in this embodiment, the first target identifier may be used as a positive sample of the model training, and after acquiring the first target identifier, the second target identifier may be obtained as a model from all identifiers of the multiple data sources. Negative sample of training.

Optionally, in this embodiment, the foregoing apparatus further includes:

a training module configured to train the prediction model according to the first target identifier and the second target identifier;

The seventh obtaining module is configured to obtain, to be pushed, a to-be-pushed identifier for the to-be-pushed resource from the identifiers included in the plurality of data sources according to the prediction model;

The push module is configured to push the to-be-pushed resource to the to-be-pushed identifier.

Optionally, in this embodiment, the acquired first target identifier and the second target identifier may be used to perform training of the predictive model, so that the to-be-pushed identifier obtained by the predictive model can more accurately represent the pointed operation. crowd. Thereby, the efficiency of pushing resources can be made higher.

The application environment of the embodiment of the present invention may be, but is not limited to, the application environment in the first embodiment, which is not described in this embodiment. An embodiment of the present invention provides an optional specific application example for implementing the foregoing method for obtaining an identifier.

As an optional embodiment, the method for obtaining the identifier may be, but is not limited to, applied to a scenario for acquiring an identifier as shown in FIG. 10 . The plurality of data sources provide data for the server, and the server obtains the first target identifier and the second target identifier according to the data obtained from the data source, and then performs training on the predictive model according to the first target identifier and the second target identifier, and is trained. The prediction model selects the identifier of the to-be-pushed resource from all the identifiers, and pushes the to-be-pushed resource to the filtered login client.

In an optional implementation manner, multiple data sources may include social/search/e-commerce/advertising/mobile application (application, referred to as app) and the like to use the identified user in social/search/e-commerce/ The user behavior in the field of advertising/mobile app is used as the characteristic information of the logo, and the primary selected crowd in each vertical industry is mined through the text semantics; the historical effect in the target data source is verified by the matching of the first identifier set and the same identifier in the preset identifier set. The saliency is given a preset weight and is sorted according to preset weights and frequency information (for example: user behavior frequency) and time information (for example: time decay factor) for the primary selection; by selecting the top N positions The identifier obtains the first target identifier, and the cross-validation of the historical effect is performed by matching the first target identifier with the preset target identifier, and the positive sample of the training data can be effectively selected; and the selected positive is subtracted from the active population of the large disc The sample set randomly acquires a second target identifier of the same size from the remaining set as a negative sample set. Thereby, the server obtains the first target identifier and the second target identifier.

In this embodiment, the positive and negative samples of the training data are obtained through text semantic feature mining, and the user's various user behavior characteristics in the social/search/e-commerce/advertising/mobile app domain are integrated, and then the user behavior frequency factor is adopted (ie, The above frequency information) and the behavior time decay factor (ie, the above time information), and the historical effect verification of the user on different behaviors, giving the user different behavior weight factors (ie, the above-mentioned preset weights), synthesizing the above elements, and making the user The scores (that is, the feature parameters obtained above) are sorted, and then the scores can be sorted according to the scores, and the purity of the positive samples (ie, the first target identifier) can be effectively determined, and the markers ranked in the first N positions can be freely selected as training according to needs. Positive sample of data. Thereby solving the problem that the user behavior is single and the purity of the positive sample is low.

In this embodiment, the behavior characteristics of the user in various scenarios on the Internet can be integrated, and the identifier corresponding to the user population with specific specific representation meanings can be mined, and the positive and negative samples with higher purity can be obtained through verification detection.

To achieve the above requirements, the foregoing server in this embodiment may include the following functional modules:

1) a feature representation word collection module, configured to define a feature representation word (corresponding to the preset feature word) according to a feature of the identifier corresponding to the specific population that needs to be filtered, which includes a positive representation word (corresponding to the first feature described above) Word) and negative representation words (equivalent to the above second feature words), wherein the positive representation words, that is, the keywords in the popular sense, the negative representation words, ie the filter words (filter_words), the negative representation words The function is to denoise, that is, to remove some of the multi-word spliced noise, so that the positive representation words can better represent our characteristic population.

2) The user multiple behavior feature fusion module is set to be refined by the user in various behaviors in the fields of social/search/e-commerce/advertising/mobile app (user identification-character representation string-time information-frequency information) ) These key elements.

3) The pattern matching module is configured to: according to the feature representation words in the feature representation word collection module, the user multiple behavior data in the user multiple behavior feature fusion module (user identification-feature representation string-time information-frequency information) In the pattern matching method, the user identifier containing the positive representation word but not the negative representation word is searched for as the primary selection identifier.

4) The user scoring module is set to score the primary selection identifier in the pattern matching module (ie, acquire the feature parameter), and the scoring involves two parts, one part is to calculate the preset weight of the data source, and the part is fine. Within each data source, calculate the behavior score of each primary identifier; where weight is calculated, there are two ways. First, the data source is divided into population packets, and the first identifier set and the preset identifier set are used. The matching of the same identifier respectively verifies the saliency of the crowd package on a single target data source, and assigns the preset weight of the current data source according to the relative value of the saliency; the other way is through the model training method, such as the LR method. The training obtains the final weight of the final data source. For example, first assign an initial weight value to each data source, and then train each data source as its feature according to the primary selected small-scale positive and negative samples, and finally iterate. After convergence, the model can spit out the preset weights of each data source.

After the data source preset weight is determined, each initial identifier is scored according to the following formula:

Where source represents the data source, there are n data sources; weight represents the preset weight on each data source; time is time information, in this example, abs (user behavior occurrence time - current Mining time), that is, the absolute value of the behavior time difference, as the user behavior time decay parameter, that is, the closer the behavior occurs to the current time, the larger the score is, the farther from the current time, the smaller the score; the action is The frequency information is used to represent the frequency of the user identification. Here, the sigmoid function is taken and normalized. The more the behavior frequency is, the higher the score is.

5) The positive and negative sample selection module is set to sort the rankings of the first N people according to the ranking of the primary selection group in the user scoring module, and select the identifiers ranked in the first N digits (the value of the N value may be different according to the orientation identifier to be mined, and the characteristic parameters) In the identification of the quantity distribution, and freely set), after the selection, the first N digits of the identification is a positive sample, the positive sample set is excluded from the identification of the active user of the large disk, and the same positive sample is selected from the remaining sets. The size of the population is identified as a negative sample.

Through the mining of text semantic features, obtaining positive and negative samples of training data can effectively avoid the problem that the seed population of the usual sense is too small, which leads to the problem that the model training characteristics are not obvious. At the same time, it can be used to measure by historical effect test and user behavior scoring. The quality of the sample improves the accuracy of the sample selection.

According to an embodiment of the present invention, there is further provided an electronic device for implementing the method for acquiring the above identifier. As shown in FIG. 11, the electronic device may include: one or more (only one shown in the figure) processor 201 The memory 203, and the transmission device 205, as shown in FIG. 11, may further include an input and output device 207.

The memory 203 can be used to store a computer program and a module, such as the method for acquiring the identifier and the program instruction/module corresponding to the device in the embodiment of the present invention. The processor 201 is configured to run the software program and the module stored in the memory 203. , thereby performing various functional applications and data processing, that is, implementing the above data loading method. Memory 203 can include high speed random access memory, and can also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory. In some examples, memory 203 can further include memory remotely located relative to processor 201, which can be connected to the terminal over a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The above described transmission device 205 is used to receive or transmit data via a network, and can also be used for data transmission between the processor and the memory. Specific examples of the above network may include a wired network and a wireless network. In one example, the transmission device 205 includes a Network Interface Controller (NIC) that can be connected to other network devices and routers via a network cable to communicate with the Internet or a local area network. In one example, the transmission device 205 is a Radio Frequency (RF) module for communicating with the Internet wirelessly.

Wherein, the memory 203 is used to store an application.

The processor 201 may call the application stored in the memory 203 through the transmission device 205 to perform the steps of: acquiring an identifier corresponding to the predetermined operation from the plurality of data sources, wherein the target data source included in the plurality of data sources The account corresponding to the identifier and the predetermined operation performed by the account are recorded; the initial identifier is obtained from the identifier according to the feature information of the identifier and the preset feature word, where the feature information is used by Determining a feature of the predetermined operation; determining a feature parameter of the initial identifier according to the preset weight and the feature information, wherein the preset weight corresponds to the target data source, and the preset weight is used to indicate The frequency of the predetermined operation is performed by the account in the target data source, the feature parameter is used to indicate the frequency at which the initial identifier performs the predetermined operation, and the first target identifier is obtained from the initial identifier, where The first target identifier is a set of identifiers in the initial identifier that are higher than a preset parameter.

The processor 201 is further configured to: acquire the first feature word and the second feature word, wherein the preset feature word includes the first feature word and the second feature word; from the identifier Obtaining the initial identifier, where the feature information corresponding to the initial identifier carries the first feature word and does not carry the second feature word.

The processor 201 is further configured to: perform the step of: acquiring the preset weight, wherein a greater value of the preset weight indicates that a higher frequency of an account in the target data source performing the predetermined operation is performed; Obtaining time information and frequency information in the feature information, wherein the time information is used to indicate a time when the identifier performs the predetermined operation, and the frequency information is used to indicate the frequency at which the identifier performs the predetermined operation; The preset weight, the time information, and the frequency information determine the feature parameter, wherein a greater value of the feature parameter indicates a higher frequency at which the initial identifier performs the predetermined operation.

The processor 201 is further configured to: perform: acquiring a proportion of an account number of the target data source that performs the predetermined operation in all accounts included in the target data source; The target data source allocates the preset weight, wherein the predetermined weight of the data source allocated by the data source is larger; and the number of the same identifier in the first identifier set and the preset identifier set is obtained, where The first identifier set is a set of identifiers included in one of the target data sources in the initial identifier; and a ratio between the quantity and the number of identifiers in the first identifier set is the target data source. The preset weight is allocated, wherein the predetermined weight of the data source allocated by the larger ratio is larger.

The processor 201 is further configured to: calculate a product of the initial identifier corresponding to the time information and the frequency information in each of the target data sources; and calculate the product according to the preset weight The weighted sum is obtained to obtain the characteristic parameter.

The processor 201 is further configured to: obtain information for indicating a feature of the predetermined operation from the predetermined operation corresponding to the identifier, wherein the information for indicating a feature of the predetermined operation And including: the feature word corresponding to the predetermined operation, the time information and the frequency information; storing the feature word, the time information, and the frequency information into a preset format to obtain the feature information.

The processor 201 is further configured to perform one of the following steps: arranging the initial identifiers according to the feature parameters from high to low; and selecting the first target identifier from the aligned identifiers, where the A target identifier includes an identifier of the top N bits in the aligned identifiers; and the first target identifier whose value of the feature parameter is greater than or equal to a preset value is obtained from the initial identifier.

The processor 201 is further configured to: perform: matching the first target identifier with a preset target identifier; and determining, if the first target identifier and the preset target identifier are successfully matched, The first target identifier is a required identifier; if the first target identifier and the preset target identifier are unsuccessful, the first target identifier is re-acquired.

The processor 201 is further configured to: determine whether the first target identifier and the preset target identifier include the same identifier greater than or equal to a preset number; and determine the first target identifier and the If the preset target identifier includes the same identifier that is greater than or equal to the preset number, the first target identifier is determined to be successfully matched with the preset target identifier.

The processor 201 is further configured to: obtain an identifier corresponding to an account that is included in the multiple data sources, and randomly obtain, in addition to the first target identifier, an identifier corresponding to an account that is included in the multiple data sources. And the identifier of the second target identifier is obtained, wherein the number of the identifiers included in the second target identifier is the same as the number of the identifiers included in the first target identifier.

The processor 201 is further configured to: perform a training prediction model according to the first target identifier and the second target identifier; and obtain, according to the prediction model, an identifier to be pushed from an identifier included by the multiple data sources. The identifier to be pushed is pushed; the to-be-pushed resource is pushed to the to-be-pushed identifier.

An embodiment of the present invention provides a solution for obtaining an identifier. Obtaining an identifier corresponding to the predetermined operation from the plurality of data sources, wherein the target data source included in the plurality of data sources records the predetermined operation performed by the account and the account corresponding to the identifier; and according to the characteristic information of the identifier and the pre- The feature word is obtained from the identifier, wherein the feature information is used to represent the feature of the predetermined operation; the feature parameter of the initial identifier is determined according to the preset weight and the feature information, wherein the preset weight corresponds to the target data source, and the preset weight is a frequency used to indicate that an account in the target data source performs a predetermined operation, the feature parameter is used to indicate a frequency at which the initial identifier performs a predetermined operation, and the first target identifier is obtained from the initial identifier, where the first target identifier is a characteristic parameter in the initial identifier A collection of identities that are higher than the preset parameters. That is to say, in the target data source, the account corresponding to the identifier and the predetermined operation performed by the account are recorded, and the identifier corresponding to the predetermined operation is obtained, so that the obtaining path of the identifier is more extensive, and the logo size is avoided from a single user log. If the identifier of the acquired identifier is biased, the initial identifier is initially filtered according to the feature information of the identifier and the preset feature word, and the feature identifier is determined according to the preset weight and the feature information to indicate the initial identifier execution. The frequency of the predetermined operation is then obtained from the initial identifier, the first target flag whose feature parameter is higher than the preset parameter, so that the identifier included in the first target identifier is an identifier that performs a predetermined operation frequency, thereby improving the acquisition. The accuracy of the identification of the training further overcomes the problem of low accuracy in obtaining the identification for training in the related art.

For example, the specific examples in this embodiment may refer to the examples described in the foregoing embodiments, and details are not described herein again.

A person skilled in the art can understand that the structure shown in FIG. 11 is merely illustrative, and the electronic device can be a smart phone (such as an Android mobile phone, an iOS mobile phone, etc.), a tablet computer, a palmtop computer, and a mobile Internet device (MID). Terminal equipment such as PAD. FIG. 11 does not limit the structure of the above electronic device. For example, the electronic device may also include more or fewer components (such as a network interface, display device, etc.) than shown in FIG. 11, or have a different configuration than that shown in FIG.

A person of ordinary skill in the art may understand that all or part of the steps of the foregoing embodiments may be completed by a program to instruct terminal device related hardware, and the program may be stored in a computer readable storage medium, and the storage medium may be Including: flash drive, read-only memory (ROM), random access memory (Random Access Memory, RAM), disk or optical disc.

Embodiments of the present invention also provide a storage medium. Optionally, in this embodiment, the foregoing storage medium may be located in at least one of the plurality of network devices in the network.

Optionally, in the present embodiment, the storage medium is arranged to store program code for performing the following steps:

S1, the identifier corresponding to the predetermined operation is obtained from the plurality of data sources, wherein the target data source included in the plurality of data sources records an account and a predetermined operation performed by the account;

S2. Acquire an initial identifier from the identifier according to the identifier information and the preset feature word, where the feature information is used to indicate a feature of the predetermined operation;

S3, determining a feature parameter of the initial identifier according to the preset weight and the feature information, where the preset weight corresponds to the target data source, and the preset weight is used to indicate a frequency at which the account in the target data source performs a predetermined operation, and the feature parameter is used to indicate Initially identifying the frequency at which the predetermined operation is performed;

S4: Acquire a first target identifier from the initial identifier, where the first target identifier is a set of identifiers in the initial identifier that have a feature parameter higher than a preset parameter.

Optionally, the storage medium is further arranged to store program code for performing the following steps:

S1: acquiring a first feature word and a second feature word, where the preset feature word includes a first feature word and a second feature word;

S2. The initial identifier is obtained from the identifier, where the feature information corresponding to the initial identifier carries the first feature word and does not carry the second feature word.

Optionally, the storage medium is further configured to store program code for performing the following steps: obtaining a preset weight, wherein a larger value of the preset weight indicates that the account in the target data source has a higher frequency of performing the predetermined operation; The time information is used to obtain the time information and the frequency information, wherein the time information is used to indicate the time when the predetermined operation is performed, the frequency information is used to indicate the frequency at which the identification performs the predetermined operation, and the characteristic parameter is determined according to the preset weight, the time information, and the frequency information. Wherein, the larger the value of the feature parameter is, the higher the frequency at which the initial identifier performs the predetermined operation.

Optionally, the storage medium is further configured to store program code for performing the following steps: acquiring a proportion of an account in the target data source that performs a predetermined operation in all accounts included in the target data source; and targeting the target data according to the ratio The source allocation preset weight, wherein the larger the proportion, the greater the preset weight of the data source allocation; or the number of the same identifier in the first identifier set and the preset identifier set, wherein the first identifier set is the initial identifier a set of identifiers included in a target data source; a preset weight is assigned to the target data source according to a ratio between the quantity and the number identified in the first identifier set, wherein the larger the ratio, the more the preset weight of the data source is assigned Big.

Optionally, the storage medium is further configured to store program code for performing the steps of: calculating a product of the initial identification of the corresponding time information and frequency information in each target data source; calculating a weighted sum of the products according to the preset weight, Get the characteristic parameters.

Optionally, the storage medium is further configured to store program code for: obtaining information for indicating a feature of the predetermined operation from the predetermined operation corresponding to the identification, wherein the information for indicating the feature of the predetermined operation comprises : Feature words, time information and frequency information corresponding to the predetermined operation; storing the feature words, time information and frequency information into a preset format to obtain feature information.

Optionally, the storage medium is further configured to store program code for performing the following steps: arranging the initial identifiers according to the feature parameters from high to low; selecting the first target identifier from the aligned identifiers, wherein the first The target identifier includes an identifier of the top N bits in the aligned identifiers; or, the first target identifier whose value of the feature parameter is greater than or equal to the preset value is obtained from the initial identifier.

Optionally, the storage medium is further configured to store program code for performing the following steps: matching the first target identifier with the preset target identifier; and determining that the first target identifier matches the preset target identifier, determining The first target identifier is the required identifier; if the first target identifier and the preset target identifier are unsuccessful, the first target identifier is re-acquired.

Optionally, the storage medium is further configured to store program code for performing the following steps: determining whether the first target identifier and the preset target identifier include the same identifier greater than or equal to the preset number; and determining the first target identifier In the case that the preset target identifier includes the same identifier that is greater than or equal to the preset number, it is determined that the first target identifier and the preset target identifier match successfully.

Optionally, the storage medium is further configured to store program code for performing the following steps: acquiring an identifier corresponding to the account number included in the plurality of data sources; randomly obtaining the identifier corresponding to the account number included in the plurality of data sources An identifier other than the target identifier obtains a second target identifier, wherein the number of the identifiers included in the second target identifier is the same as the number of the identifiers included in the first target identifier.

Optionally, the storage medium is further configured to store program code for performing the following steps: training the prediction model according to the first target identifier and the second target identifier; and selecting the resource to be pushed from the identifiers included in the plurality of data sources according to the prediction model Acquire the to-be-pushed identifier; push the to-be-pushed resource to the to-be-pushed identifier.

Optionally, in this embodiment, the foregoing storage medium may include, but not limited to, a USB flash drive, a Read-Only Memory (ROM), a Random Access Memory (RAM), a mobile hard disk, and a magnetic memory. A variety of media that can store program code, such as a disc or a disc.

For example, the specific examples in this embodiment may refer to the examples described in Embodiment 1 and Embodiment 2, and details are not described herein again.

The serial numbers of the embodiments of the present invention are merely for the description, and do not represent the advantages and disadvantages of the embodiments.

The integrated unit in the above embodiment, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in the above-described computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product in the form of a software product, or the whole or part of the technical solution, which is stored in a storage medium, including The instructions are used to cause one or more computer devices (which may be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention.

In the above-mentioned embodiments of the present invention, the descriptions of the various embodiments are different, and the parts that are not detailed in a certain embodiment can be referred to the related descriptions of other embodiments.

In the several embodiments provided by the present application, it should be understood that the disclosed client may be implemented in other manners. The device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner. For example, multiple units or components may be combined or may be Integrate into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, unit or module, and may be electrical or otherwise.

The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

The above description is only a preferred embodiment of the present invention, and it should be noted that those skilled in the art can also make several improvements and retouchings without departing from the principles of the present invention. It should be considered as the scope of protection of the present invention.

Industrial applicability:

According to the above description, the present invention records the account corresponding to the identifier and the predetermined operation performed by the account in the target data source, and obtains the identifier corresponding to the predetermined operation, so that the acquisition path of the identifier is more extensive, and the single user log is avoided. Obtaining the problem that the acquired identifier is biased due to the small size of the identifier, and then initially screening the initial identifier according to the feature information of the identifier and the preset feature word, and determining the feature parameter according to the preset weight and the feature information to represent the feature identifier Initially identifying the frequency of performing the predetermined operation, and then acquiring, from the initial identifier, the first target flag whose feature parameter is higher than the preset parameter, so that the identifier included in the first target identifier is an identifier that performs a predetermined operation frequency, thereby improving The accuracy of the identification for training is obtained, thereby overcoming the problem of low accuracy in obtaining the identification for training in the related art.

Claims (17)

A method for obtaining an identifier, comprising: Acquiring an identifier corresponding to the predetermined operation from the plurality of data sources, wherein the target data source corresponding to the identifier and the predetermined operation performed by the account account are recorded in the target data source included in the plurality of data sources; Obtaining an initial identifier from the identifier according to the feature information of the identifier and a preset feature word, where the feature information is used to indicate a feature of the predetermined operation; Determining, according to the preset weight and the feature information, a feature parameter of the initial identifier, where the preset weight is corresponding to the target data source, and the preset weight is used to indicate an account execution in the target data source a frequency of the predetermined operation, the feature parameter is used to indicate a frequency at which the initial identification performs the predetermined operation; Obtaining a first target identifier from the initial identifier, where the first target identifier is a set of identifiers in the initial identifier that are higher than a preset parameter. The method according to claim 1, wherein the obtaining the initial identifier from the identifier according to the identifier information and the preset feature word comprises: Obtaining a first feature word and a second feature word, wherein the preset feature word includes the first feature word and the second feature word; Obtaining the initial identifier from the identifier, where the feature information corresponding to the initial identifier carries the first feature word and does not carry the second feature word. The method according to claim 1, wherein the determining the characteristic parameter of the initial identifier according to the preset weight and the feature information comprises: Obtaining the preset weight, wherein a value of the preset weight indicates that a frequency of performing an operation of the account in the target data source is higher; Obtaining time information and frequency information from the feature information, wherein the time information is used to indicate a time when the identifier performs the predetermined operation, and the frequency information is used to indicate a frequency at which the identifier performs the predetermined operation ; Determining the feature parameter according to the preset weight, the time information, and the frequency information, wherein a greater value of the feature parameter indicates that a frequency of the initial identifier performing the predetermined operation is higher. The method of claim 3, wherein obtaining the preset weight comprises one of the following: Obtaining a proportion of an account number of the target data source that performs the predetermined operation in all the account accounts included in the target data source; and assigning the preset weight to the target data source according to the ratio, where The greater the proportion, the greater the predetermined weight assigned by the data source; Obtaining, by the first identifier set, a quantity of the same identifier in the preset identifier set, where the first identifier set is a set of identifiers included in one target data source in the initial identifier; The ratio between the number of identifiers in the first identifier set is the target data source, and the preset weight is allocated by the data source with the larger ratio. The method according to claim 3, wherein determining the feature parameters according to the preset weight, the time information, and the frequency information comprises: Calculating a product of the time information corresponding to the initial identifier in each of the target data sources and the frequency information; Calculating a weighted sum of the products according to the preset weights to obtain the feature parameters. The method according to claim 3, wherein the method further comprises: before acquiring the initial identifier from the identifier according to the identifierd feature information and the preset feature word, the method further comprising: Acquiring information for indicating a feature of the predetermined operation from the predetermined operation corresponding to the identifier, wherein the information for indicating a feature of the predetermined operation includes: a feature word corresponding to the predetermined operation, The time information and the frequency information; And storing the feature word, the time information, and the frequency information into a preset format to obtain the feature information. The method of claim 1, wherein obtaining the first target identifier from the initial identifier comprises one of: Arranging the initial identifiers according to the feature parameters from high to low; selecting the first target identifier from the aligned identifiers, wherein the first target identifiers are included in the aligned identifiers N-digit identification; Obtaining, from the initial identifier, the first target identifier that the value of the feature parameter is greater than or equal to a preset value. The method according to any one of claims 1 to 7, wherein after the obtaining the first target identifier from the initial identifier, the method further comprises: Matching the first target identifier with a preset target identifier; If the first target identifier is successfully matched with the preset target identifier, the first target identifier is determined to be a required identifier; and the first target identifier matches the preset target identifier. In case of success, the first target identifier is re-acquired. The method of claim 8, wherein matching the first target identifier with the preset target identifier comprises: Determining whether the first target identifier and the preset target identifier include the same identifier that is greater than or equal to a preset number; If it is determined that the first target identifier and the preset target identifier include the same identifier that is greater than or equal to the preset number, the first target identifier is determined to be successfully matched with the preset target identifier. The method according to any one of claims 1 to 7, wherein after the obtaining the first target identifier from the initial identifier, the method further comprises: Obtaining an identifier corresponding to an account included in the multiple data sources; Obtaining an identifier other than the first target identifier from the identifier corresponding to the account that is included in the multiple data sources, to obtain a second target identifier, where the number of identifiers included in the second target identifier is The number of the identifiers included in the first target identifier is the same. The method according to claim 10, wherein the identifiers other than the first target identifier are randomly obtained from the identifiers corresponding to the account numbers included in the plurality of data sources, and after the second target identifier is obtained, the The method also includes: Training a prediction model according to the first target identifier and the second target identifier; Obtaining, to be pushed, an identifier to be pushed for the to-be-pushed resource from the identifiers included in the multiple data sources according to the prediction model; Pushing the to-be-pushed resource to the to-be-pushed identifier. An apparatus for obtaining identification, comprising one or more processors, and one or more memories storing program units, wherein the program units are executed by the processor, the program units comprising: The first obtaining module is configured to obtain an identifier corresponding to the predetermined operation from the plurality of data sources, where the account corresponding to the identifier and the account number are recorded in the target data source included in the plurality of data sources Executed operation; a second obtaining module, configured to obtain an initial identifier from the identifier according to the identifier information and the preset feature word, where the feature information is used to indicate a feature of the predetermined operation; a determining module, configured to determine a feature parameter of the initial identifier according to the preset weight and the feature information, where the preset weight corresponds to the target data source, and the preset weight is used to indicate the target An account number in the data source performs a frequency of the predetermined operation, the feature parameter is used to indicate a frequency at which the initial identifier performs the predetermined operation; And a third acquiring module, configured to obtain a first target identifier from the initial identifier, where the first target identifier is a set of identifiers in the initial identifier that are higher than a preset parameter. The apparatus of claim 12, wherein the second acquisition module comprises: a first acquiring unit, configured to acquire the first feature word and the second feature word, wherein the preset feature word includes the first feature word and the second feature word; The second acquiring unit is configured to obtain the initial identifier from the identifier, where the feature information corresponding to the initial identifier carries the first feature word and does not carry the second feature word. The apparatus of claim 12, wherein the determining module comprises: The third obtaining unit is configured to acquire the preset weight, wherein a larger value of the preset weight indicates that a frequency of performing an operation of the account in the target data source is higher; a fourth obtaining unit, configured to obtain time information and frequency information from the feature information, where the time information is used to indicate a time when the identifier performs the predetermined operation, and the frequency information is used to indicate the Identifying a frequency at which the predetermined operation is performed; a determining unit, configured to determine the feature parameter according to the preset weight, the time information, and the frequency information, wherein a greater value of the feature parameter indicates a frequency at which the initial identifier performs the predetermined operation The higher. The apparatus of claim 12, wherein the third acquisition module comprises one of: a processing unit, configured to: arrange the initial identifier according to the feature parameter from high to low; select the first target identifier from the aligned identifiers, wherein the first target identifier is included in the array The identifier of the top N in the logo; The fifth obtaining unit is configured to obtain, from the initial identifier, the first target identifier that the value of the feature parameter is greater than or equal to a preset value. A storage medium, wherein a computer program is stored in the storage medium, the computer program being arranged to execute the method of any one of claims 1 to 11 at runtime. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, the processor being arranged to perform the method of any one of claims 1 to 11 by the computer program .

Images