HK1166393A - Method and apparatus for ranking search results - Google Patents

Description

Search sorting method and device

Technical Field

The present application relates to the field of computer application technologies, and in particular, to a search ranking method and apparatus.

Background

The online transaction platform provides a platform for ensuring the safety of online transactions, and the online transactions are commodity or service transactions conducted by buyers and sellers through the transaction platform. The online trading platform provides various products, and users search according to needs to find needed commodities from search results. The online trading platform can provide different search channels for users according to different application markets, such as: the system comprises a main search channel, a retail search channel, a category search channel, a purchase search channel, a hot-sales search channel and the like, wherein each search channel has the requirements of respective product information and sequencing.

In the prior art, different search channels correspond to different search servers, each search server provides corresponding product information for a corresponding search channel, when a buyer user needs to search for a product in a certain search channel, the search server corresponding to the search channel receives query information submitted by the user and clicks page information entering the search channel, then the search server corresponding to the search channel searches all product information related to a keyword input by the user from a storage server in which the product information is stored, then the related product information is ranked, in the ranking process, product features of the product information need to be extracted respectively under different search channels, product feature scores are calculated, and ranking is performed according to the score condition.

As can be seen from the above process, each search channel corresponds to its own search server, and the respective search server searches the product information of the seller user in the corresponding search channel and performs feature extraction on the searched product information, and in fact, many product information in each search channel is duplicated, for example, a minor search channel is a search channel providing consumer markets such as clothing and small goods, and a main search channel is a search channel providing all product information, which necessarily includes the product information in the minor search channel, but according to the search ranking method of the prior art, the duplicated product information needs to be duplicated in different search servers corresponding to different search channels, which causes waste of information resources and hardware resources, and because many product information are duplicated in different search channels, when sorting products, the search servers corresponding to different search channels need to repeatedly extract product features and respectively calculate, which also causes waste of information resources and hardware resources, thereby affecting the processing performance of the servers. In addition, in the prior art, the same sorting algorithm is adopted for different search channels in the sorting process, so that the unique characteristic information in the different search channels cannot be fully utilized to carry out targeted sorting, the sorting result information in each search channel cannot well accord with the product characteristics of the channel, and the searching requirements of users cannot be met.

In summary, one of the technical problems that needs to be urgently solved by those skilled in the art is: how to creatively provide a search sequencing method to solve the problem of reduced server processing performance caused by waste of information resources and hardware resources in the prior art.

Disclosure of Invention

The technical problem to be solved by the application is to provide a search sorting method, which solves the problem of decreased server processing performance caused by waste of information resources and hardware resources, and further enables the sorting method in each search channel to be performed based on the feature information in the channel, so that the product features in the channel can be reflected, and further the sorting result information can meet the search requirements of users.

The application also provides a search sequencing device for ensuring that the method is realized in practice and is about to be applied.

In order to solve the above problem, the present application discloses a search ranking method, including:

the method comprises the steps that a search server receives query information and search channel information of a user, wherein the query information is query keywords or clicked category information;

the search server inquires related product information related to the inquired information in the search channel;

the search server respectively calculates a relevance score and a product information characteristic score of the related product information, wherein the relevance score represents the matching degree of the query information and the product information;

the search server ranks the related product information according to the relevance score and the product information feature score.

Preferably, the searching server querying the related product information related to the query information in the search channel includes:

the search server acquires all product information of the search channel according to the search channel information;

and the search server inquires related product information related to the inquired information in all the product information according to the inquired information.

Preferably, when the query information is a query keyword, the search server queries related product information related to the query information from all the product information according to the query information, and the query information includes:

the search server obtains at least one key field according to the query keyword;

and inquiring related product information matched with the at least one key field in all the product information.

Preferably, the search server calculates the relevance score of the related product information, specifically:

and the search server calculates the relevance score of the related product information according to a pre-trained relevance ranking model corresponding to the search channel.

Preferably, when the query information is clicked category information, the search server queries related product information related to the query information in the product information according to the query information, specifically: and the server inquires related product information matched with the category information in all the product information according to the clicked category information.

Preferably, the search server calculates the relevance score of the related product information, specifically:

the search server calculates a relevance score according to a degree of reliability of the product information and the category, the degree of reliability representing a degree of proximity of the product information description to the category.

Preferably, the calculating, by the search server, the product information feature score of the related product information includes:

and the search server inquires whether the product information characteristics of the related product information are stored or not, if so, directly calculates the product information characteristic score according to a pre-trained product characteristic sorting model corresponding to the search channel, otherwise, extracts the product information characteristics of each piece of related product information, and calculates the extracted product information characteristic score according to the pre-trained product characteristic sorting model corresponding to the search channel.

Preferably, after extracting the product features of each piece of related product information, the method further includes:

and the search server stores the product information characteristics of each piece of extracted relevant product information.

The application discloses search sequencing device, the device is applied to on the search server of online trading platform, and the device includes:

the receiving module is used for receiving query information and search channel information of a user, wherein the query information is query keywords or clicked category information;

the query module is used for querying related product information related to the query information in the search channel;

the calculation module is used for respectively calculating the correlation score and the product information characteristic score of the related product information; the relevance score represents the matching degree of the query information and the product information;

and the sorting module is used for sorting the related product information according to the relevance score and the product information characteristic score.

Preferably, the query module includes:

the information acquisition module is used for acquiring all product information of the search channel according to the search channel information;

and the information query module is used for querying the related product information related to the query information in all the product information according to the query information.

Preferably, when the query information is a query keyword, the information query module includes:

the field acquisition module is used for acquiring at least one key field according to the query keyword;

and the product query module is used for querying the related product information matched with the at least one key field in all the product information.

Preferably, the calculation module is specifically configured to calculate a relevance score for the related product information according to a pre-trained ranking model corresponding to a search channel; and the number of the first and second groups,

and the product characteristic score is calculated according to the relevant product information and a pre-trained product characteristic sorting model corresponding to the search channel.

Preferably, when the query information is clicked category information, the information query module is specifically configured to query, according to the clicked category information of the user, related product information related to the category information in all the product information.

Preferably, the calculation module is specifically configured to calculate a correlation score according to a reliability degree of the product information and the category, where the reliability degree represents a degree of closeness of the product information description to the category; and the number of the first and second groups,

and the product characteristic score is calculated according to the relevant product information and a pre-trained product characteristic sorting model corresponding to the search channel.

Preferably, the computing module includes:

the query judging module is used for querying whether the product characteristics of the related product information are stored or not;

the characteristic extraction module is used for extracting the product characteristics of each piece of related product information when the result of the query judgment module is negative;

the characteristic calculation module is used for directly calculating the characteristic score of the product information according to a pre-trained product characteristic sorting model corresponding to the search channel when the result of the query judgment module is positive; or, when the result of the query judgment module is negative, calculating the product information feature score extracted by the feature extraction module according to a pre-trained product feature sorting model corresponding to the search channel.

Preferably, the apparatus further comprises:

and the characteristic storage module is used for storing the product information characteristics of each piece of related product information extracted by the characteristic extraction module.

Compared with the prior art, the method has the following advantages:

in the method, the unified search server is arranged aiming at different search channels, the search server receives the query information of the user in a unified mode, then the corresponding product information is obtained by querying from the storage server in which the product information is stored, repeated copying of the product information is avoided, sequencing operation is carried out by the same search server, product characteristics are stored, repeated extraction of the product information characteristics is avoided, therefore, sharing of information of different search channels is achieved, information resources and hardware resources are saved, and the problem that the processing performance of the server is reduced is solved.

Of course, it is not necessary for any product to achieve all of the above-described advantages at the same time for the practice of the present application.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive labor.

FIG. 1 is a flow chart of embodiment 1 of a search ranking method of the present application;

FIG. 2 is a flow chart of embodiment 2 of a search ranking method of the present application;

FIG. 3 is a flow chart of embodiment 3 of a search ranking method of the present application;

fig. 4 is a block diagram of a search ranking apparatus according to embodiment 1 of the present application;

fig. 5 is a block diagram of a search ranking apparatus according to embodiment 2 of the present application;

FIG. 6 is a block diagram of the structure of an information query module in the present application;

fig. 7 is a block diagram of a computing module according to the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The application is operational with numerous general purpose or special purpose computing device environments or configurations. For example: personal computers, server computers, hand-held or portable devices, tablet-type devices, multi-processor apparatus, distributed computing environments that include any of the above devices or equipment, and the like.

The application may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The application may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

One of the main ideas of the present application may include that the search server determines a search channel according to query information of a user, matches related product information related to the query of the user in product information corresponding to the search channel, and then ranks the related products. In the method, the unified search server is arranged for different search channels, the search server receives the query information of the user in a unified mode, then the corresponding product information is obtained by querying from the storage server in which the product information is stored, and the product information is sequenced in the search server, so that the information sharing of the different search channels is realized, the information resources and the hardware resources are saved, and the problem of the reduction of the processing performance of the server is solved.

Referring to fig. 1, a flowchart of an embodiment 1 of a search ranking method according to the present application is shown, which may include the following steps:

step 101: the search server receives query information and search channel information of a user, wherein the query information is query keywords or clicked category information.

When a user searches, the user clicks the link set for each search channel on the main page to enter the page of a certain search channel, and the user can input a query keyword or click the category link provided by the page of the search channel to query. For example, the user needs to search for "mobile phone", and may input "mobile phone" in the search box of the search channel page, or find the "mobile phone" category in the category link provided in the search channel page and click directly.

In the embodiment of the application, a unified search server is set for different search channels, the search server receives query information and search channel information of a user in a unified manner, and the search server can judge which channel the query of the user comes from when the user clicks a search channel page according to the corresponding relation between the search channel page information and the search channel.

Step 102: and the search server inquires related product information related to the inquired information in the search channel.

The search server judges a search channel corresponding to user query according to the click page information of the user, all product information corresponding to the search channel is searched and obtained from a storage server storing the product information, for example, a random search channel is a search channel opened for special transaction of related clothes, small commodities and other consumer goods in the product information of a seller, the user clicks to enter the search channel, the server can obtain information of all consumer goods from the product information of the seller, and then related product information matched with the query information of the user is obtained from all consumer goods through query. In specific implementation, the method and the device are not limited to be applied to searching product information on an online transaction platform, and can also be applied to searching various information on various portal websites, search engine websites and the like.

Step 103: the search server calculates a relevance score and a product information feature score of the related product information, respectively.

The relevance score refers to the degree of matching of the user query information with the related product information. Taking the query keyword information input by the user as an example, for example, when the user searches for "nokia handset", one of the pieces of matching information is given as table 1.

Referring to table 1, the correlation score calculation process may be exemplified as follows:

the keyword input by the user and the piece of product information are matched, but the interval between the keywords is '5230', so that the matching characteristic a is 0.95; according to historical click data statistics, a click score b under a category mobile phone is 1, a keyword matches a brand in attribute Nokia, an attribute matching feature score c is 1, information attribute description and detailed description fields relate to the content of the mobile phone, an overall matching feature score d is 1, and values a, b, c and d are substituted into a relevance ranking model to obtain a final relevance score.

It should be noted that, according to the user query keyword, the search server queries all related products related to the query keyword in the product information in the search channel, table 1 only gives one piece of product information, and the calculation process is only described briefly, and those skilled in the art should understand that the matching features of the product information are not limited to a, b, c, and d. The calculation of the correlation of other product information is not listed here.

TABLE 1

The product information features characterize the quality status of the product, for example, a piece of product information may include three dimensions of features: an information dimension feature, a seller dimension feature, and a market dimension feature.

As shown in table 2, the information dimension feature may include an attribute filling feature of the information, an information publishing time feature, and an information description feature, and the server characterizes the features by some values according to various aspects of the information dimension feature, for example, the information description is detailed, and the server characterizes the feature by 0.8 in a similar manner, for example, the information publishing time is within 3 days, and then characterizes the feature by a value of 1. The seller dimension characteristics can comprise liveness characteristics, credit guarantee characteristics, contribution condition characteristics to websites and the like; market dimension characteristics may include aspects such as buyer feedback optimization and market trading maximization.

Likewise, the search server may numerically characterize the features in these dimensions based on their aspects. Specifically, refer to table 2, which is not repeated herein, and it should be noted that table 2 only describes some features thereof.

TABLE 2

And the product information characteristic score is calculated by substituting score values of all the dimensional characteristics into a product characteristic sorting model.

In the application, the feature information of each dimension of the product may be extracted from information filled in when the product information is published, for example, the publishing time feature of the information, or/and obtained from the recorded historical behavior data of the seller, for example, the credit guarantee feature of the seller.

Step 104: the search server sorts the related product information according to relevance scores and product information feature scores.

The server sorts according to the relevance scores, sorts according to the product information characteristic scores in the relevance sorting results, and finally displays the sorting results on the search page. It should be noted that, in the embodiment of the present application, the obtained product information features may be stored in the search server, and when calculating the product information feature score, the product information features may be directly used. Moreover, the execution main body in this embodiment is different from the search server corresponding to only one search channel in the prior art, and the search server in this embodiment is a set of hardware and software corresponding to five search channels at the same time, which is a newly added computer system with greatly improved hardware performance and software performance. It can also be understood that, after the search server in this embodiment is subjected to performance enhancing operations such as expansion or improvement, the hardware and software systems of the search server are greatly enhanced, all product information in five search channels can be provided simultaneously in terms of hardware, and at the same time, the search server has better processing performance in terms of software.

In this embodiment, a unified search server is set for different search channels, the search server receives query information of a user in a unified manner, then corresponding product information is obtained by querying from a storage server in which the product information is stored, all the product information is obtained by querying by the same search server, repeated copying of the product information is avoided, and the same search server performs a sorting operation, and the search server can store product characteristics and avoid repeated extraction of the product characteristics, thereby realizing sharing of information of different search channels, saving information resources and hardware resources, and solving a problem of degradation of processing performance of the server.

Referring to fig. 2, a flowchart of an embodiment 2 of a search ranking method according to the present application is shown, where the embodiment 2 is a specific flowchart when query information of a user is a query keyword, and may include the following steps:

step 201: the search server receives query keyword information and search channel information of a user.

The main execution body in this embodiment is the same search server as in embodiment 1 of the present application. In this step, the search server determines which search channel the user's query comes from according to the search channel page information clicked by the user.

Step 202: and the search server acquires corresponding product information of the search channel according to the search channel information.

Step 203: and the search server inquires related product information related to the inquiry keyword in the product information according to the inquiry keyword.

And the search server obtains at least one key field according to the query key words, and queries related product information matched with the at least one key field in all the product information. For example, the keyword input by the user is "nokia handset", the key fields of the keyword include at least two key fields of "nokia" and "handset", and the search server will query at least one piece of information matched with one key field from all product information corresponding to the search channel.

Step 204: and the search server calculates the relevance score of the related product information according to a pre-trained relevance ranking model corresponding to the search channel.

The relevance score refers to the degree of matching of the user query keyword with the related product information. The search server pre-trains a relevance ranking model corresponding to the search channel aiming at different search channels. The correlation Ranking model is a linear-based regression model, and preferably, the correlation Ranking model may be an MLR (Machine Learning Ranking model), and the MLR correlation Ranking model may further be a support vector Machine model.

The process of calculating the relevance score of the related product information related to the query keyword can be found in embodiment 1, and is not described herein again.

Step 205: the search server queries whether the product information features of the related product information have been saved, if so, proceeds to step 207, and if not, proceeds to step 206.

The product information features are stored in a specific storage module feature pool in the search server, and when calculating the product information feature scores, the search server firstly inquires whether the product information features of the related product information are stored, namely whether the product information features are stored in the feature pool.

The feature pool is a storage module for storing the product information features in the search server, and is not limited to the name of the feature pool.

Step 206: the search server extracts product information characteristics of each piece of related product information.

The product information features can be obtained from the seller's historical behavior data extracted and/or recorded from the information filled in when the product information is published.

The product information features for extracting each piece of product information are actually represented in numerical form according to the condition of each dimension feature of the product. The numerical calculation process can be specifically described in example 1.

Step 207: and calculating the product information characteristic score according to a pre-trained product characteristic sorting model corresponding to the search channel.

And if the search server does not store the product information features of the related product information, calculating the extracted product information feature score according to the pre-trained product feature ranking model corresponding to the search channel.

The product information features represent the quality condition of the product information, the search server represents each feature condition of certain information according to the numerical value, and when the product feature score is calculated, the product information feature score is calculated by substituting the numerical value representing each dimension feature into a pre-trained product feature model. The product feature model is a regression model based on linearity, for example, the regression model may be a linear expression, and the server may train and learn in advance for each search channel according to the product information features to obtain a corresponding feature ranking model.

Step 208: the search server ranks the product information according to the relevance score and the product information feature score.

And the search server sorts according to the degree of the product information relevance score to obtain a relevance sorting result, and sorts according to the product information characteristic score in the relevance sorting result.

Step 209: and the search server stores the product information characteristics of each piece of extracted relevant product information.

The product information features represented by the numerical values are stored in a feature pool of the search server, and the feature pool is a storage module in the search server and used for storing the product information features. When the search server calculates the product information feature score, the search server directly selects the features from the feature pool and substitutes the features into the product feature sorting model without repeating the extraction and calculation process.

It should be noted that the operation of step 209 is not limited to the order of embodiment 2, and may be performed after or simultaneously with any one of step 206 to step 208.

After the search server carries out the sequencing, the final sequencing result can be displayed on the search page.

In this embodiment, a unified search server is set for different search channels, the search server receives query information of a user in a unified manner, then corresponding product information is obtained by querying from a storage server in which the product information is stored, all the product information is provided by the same search server, repeated copying of the product information is avoided, and the same search server performs a sorting operation. Therefore, the embodiment realizes the sharing of different search channel information, saves information resources and hardware resources, and solves the problem of the reduction of the processing performance of the search server. Furthermore, in this embodiment, for the unique feature information in each search channel, the extraction of the product information features and the score calculation process of the product information features are performed first, so that for different search channels, the repeated extraction process of the same product information features can be avoided, and the characteristics of the unique product information features can be reflected, so that the ranking results in each search channel can well meet the product characteristics of the channel, and the search requirements of the user are met.

Referring to fig. 3, a flowchart of embodiment 3 of a search ranking method according to the present invention is shown, where embodiment 3 is a specific flowchart when query information of a user is category information clicked by the user, and the specific flowchart may include:

step 301: the search server receives category information clicked by the user and search channel information.

The search server in the present embodiment is also the same as that in embodiment 1. The user clicks a page entering a certain search channel, searches needed by the user are searched in the category links provided by the search channel page, and the user clicks directly. The search server performs product information matching and product information sorting operations according to the click information of the user.

Step 302: and the search server acquires all product information of the search channel according to the search channel information.

Step 303: and the search server inquires related product information related to the category information in all the product information according to the category information clicked by the user.

And the search server inquires related product information of which the category is 'mobile phone' from the product information corresponding to the search channel according to the category information clicked by the user, for example, the user clicks the 'mobile phone' category.

Step 304: the search server calculates a relevance score for the related product information.

The search server calculates the reliability score of the seller under each category according to the categories to which the seller product information belongs, namely the matching degree of the seller product information and the categories to which the seller product information belongs.

For example, if the user clicks the category "mobile phone", the obtained related product information also has the information shown in table 1, and the product described by the product information in table 1 is "mobile phone", the reliability score of the product in the category "mobile phone" is 1, and the relevance score of the piece of information is 1.

Step 305: the search server queries whether the product characteristics of the related product information are stored, if so, proceeds to step 307, and if not, proceeds to step 306.

The product information features are stored in a specific storage module feature pool in the search server, and when calculating the product information feature scores, the search server firstly inquires whether the product information features of the related product information are stored, namely whether the product information features are stored in the feature pool.

Step 306: and extracting and storing the product characteristics of each piece of related product information.

The extracted product information features are stored in a feature pool in the search server, wherein the feature pool is a specific storage module in the search server.

Step 307: and calculating the product characteristic score according to a pre-trained product characteristic sorting model corresponding to the search channel.

Step 308: the search server ranks the product information according to the relevance score and the product information feature score.

And the search server sorts according to the degree of the product information relevance score to obtain a relevance sorting result, and sorts according to the product information characteristic score in the relevance sorting result.

Step 309: and after the search server carries out sequencing, displaying a final sequencing result on a search page.

It should be noted that the difference between the embodiment 3 and the embodiment 2 is only that the processes of querying the related product information and calculating the relevance score are different, and other steps can be referred to each other.

In the embodiment, a unified search server is arranged for different search channels, the search server receives query information of users in a unified manner, corresponding product information is queried and obtained from a storage server in which the product information is stored, all the product information is provided by the same search server, repeated copying of the product information is avoided, the same search server performs sorting operation, in the sorting process, the product information characteristics are stored when the search server extracts the product information characteristics, and when product information characteristic score calculation is performed again, the repeated extraction operation is not required, and the product information characteristic score can be directly calculated. Therefore, the sharing of different search channel information is realized, information resources and hardware resources are saved, and the problem that the processing performance of the search server is reduced is solved.

While, for purposes of simplicity of explanation, the foregoing method embodiments have been described as a series of acts or combination of acts, it will be appreciated by those skilled in the art that the present application is not limited by the order of acts or acts described, as some steps may occur in other orders or concurrently with other steps in accordance with the application. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required in this application.

Corresponding to the method provided by the above-mentioned embodiment 1 of the search ranking method in the present application, referring to fig. 4, the present application further provides an embodiment 1 of a search ranking apparatus, and in this embodiment, the apparatus may include:

the receiving module 401 is configured to receive query information and search channel information of a user, where the query information is a query keyword or clicked category information.

A query module 402, configured to query the search channel for related product information related to the query information.

A calculating module 403, configured to calculate a relevance score and a product information feature score of the related product information respectively; the relevance score represents a degree of matching of the query information and the product information.

A ranking module 404, configured to rank the related product information according to the relevance score and the product information feature score.

The apparatus described in this embodiment may be integrated into a server of a search engine, or may be connected to the search engine server separately as an entity, and in addition, when the method described in this application is implemented by software, the apparatus may be used as a function newly added to the server of the search engine, or may write a corresponding program separately, and the application does not limit the implementation manner of the method or the apparatus.

In the embodiment, the unified search server is arranged for different search channels, the search server receives the query information of the user in a unified manner, then the corresponding product information is obtained by querying from the storage server in which the product information is stored, all the product information is provided by the same search server, repeated copying of the product information is avoided, the same search server performs sequencing operation, the search server can store the product characteristics, repeated extraction of the product characteristics is avoided, therefore, sharing of information of different search channels is realized, information resources and hardware resources are saved, and the problem that the processing performance of the search server is reduced is solved.

Corresponding to the method provided by the above embodiment 2 of the search ranking method in the present application, referring to fig. 5, the present application further provides an embodiment 2 of a search ranking apparatus, and in this embodiment, the apparatus may specifically include:

the receiving module 501: for receiving query information and search channel information of a user, the

The information acquisition module 502: and the system is used for acquiring all product information of the search channel according to the search channel information.

The information query module 503: and the system is used for inquiring related product information related to the inquiry keyword in all the product information according to the inquiry keyword.

As shown in fig. 6, the information query module 503 includes:

the field acquisition module 601: for deriving at least one key field from the query keyword.

Product query module 602: the system is used for inquiring the related product information matched with the at least one key field in all the product information.

The calculation module 504: the correlation score is calculated according to a pre-trained sorting model corresponding to the search channel; and the number of the first and second groups,

and the product characteristic score is calculated according to the relevant product information and a pre-trained product characteristic sorting model corresponding to the search channel.

As shown in fig. 7, the calculation module 504 may specifically include:

the query judging module 701: product features for querying whether the related product information is stored.

Feature extraction module 702: and when the result of the query judging module is negative, extracting the product characteristics of each piece of relevant product information.

The feature calculation module 703: the product information feature score is directly calculated according to a pre-trained product feature sorting model corresponding to a search channel when the result of the query judging module is positive; or, when the result of the query judgment module is negative, calculating the product feature score extracted by the feature extraction module according to a pre-trained product feature ranking model corresponding to the search channel.

The sorting module 505: the system is used for sorting the related product information according to the relevance score and the product information characteristic score.

After the ranking is performed, the final ranking result is displayed on the search page.

Feature preservation Module 506: and the product information characteristic of each piece of related product information extracted by the characteristic extraction module is stored.

In practical applications, the feature saving module 506 may be embodied as a feature pool in the search server.

Corresponding to the method provided in embodiment 3 of the search ranking method of the present application, the present application further provides a search ranking apparatus, which is different from the search ranking apparatus of embodiment 2 only in that:

a receiving module: and the system is used for receiving the query information clicked by the user and the search channel information. The query information is category information clicked by a user.

An information inquiry module: and the system is used for inquiring related product information related to the category information in all the product information according to the category information clicked by the user.

A calculation module: specifically, the system is used for calculating the correlation score according to the reliability degree of the product information and the category, wherein the reliability degree represents the proximity degree of the product information description and the category; and the number of the first and second groups,

and the product characteristic score is calculated according to the relevant product information and a pre-trained product characteristic sorting model corresponding to the search channel.

Wherein, the calculation module may specifically include:

the query judgment module: product features for querying whether the related product information is stored.

A feature extraction module: and when the result of the query judging module is negative, extracting the product characteristics of each piece of relevant product information.

A feature calculation module: the product information feature score is directly calculated according to a pre-trained product feature sorting model corresponding to a search channel when the result of the query judging module is positive; or, when the result of the query judgment module is negative, calculating the product feature score extracted by the feature extraction module according to a pre-trained product feature ranking model corresponding to the search channel.

Other functional modules can be referred to each other, and are not described herein again.

The present application further provides a search server, where the search server includes the above search ranking device, and it should be noted that those skilled in the art can understand that the search server includes not only the above search ranking device. Compared with the search server in the prior art, the search server can simultaneously correspond to a set of hardware and software of five search channels, and is a newly added computer system with greatly improved hardware performance and software performance. It can also be understood that after the search server is subjected to performance-improving operations such as capacity expansion or improvement, the hardware and software systems of the search server are greatly improved, all product information in five search channels can be provided simultaneously in the aspect of hardware, and meanwhile, the search server also has better processing performance in the aspect of software; and after the search sorting device is integrated, the search sorting method can be realized.

It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the device-like embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.

Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. The term "comprising", without further limitation, means that the element so defined is not excluded from the group consisting of additional identical elements in the process, method, article, or apparatus that comprises the element.

For convenience of description, the above devices are described as being divided into various units by function, and are described separately. Of course, the functionality of the units may be implemented in one or more software and/or hardware when implementing the present application.

From the above description of the embodiments, it is clear to those skilled in the art that the present application can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present application may be essentially or partially implemented in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments or some parts of the embodiments of the present application.

The search ranking method and apparatus provided by the present application are introduced in detail above, and a specific example is applied in the text to explain the principle and the implementation of the present application, and the description of the above embodiment is only used to help understand the method and the core idea of the present application; meanwhile, for a person skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.

Claims (16)

1. A method of search ranking, the method comprising:

the method comprises the steps that a search server receives query information and search channel information of a user, wherein the query information is query keywords or clicked category information;

the search server inquires related product information related to the inquired information in the search channel;

the search server respectively calculates a relevance score and a product information characteristic score of the related product information, wherein the relevance score represents the matching degree of the query information and the product information;

the search server ranks the related product information according to the relevance score and the product information feature score.

2. The method of claim 1, wherein the searching server querying the search channel for related product information related to the query information comprises:

the search server acquires corresponding product information of the search channel according to the search channel information;

and the search server inquires related product information related to the inquiry information in the product information according to the inquiry information.

3. The method according to claim 2, wherein when the query information is a query keyword, the search server queries related product information related to the query information from all the product information according to the query information, and the method comprises:

the search server obtains at least one key field according to the query keyword;

and inquiring related product information matched with the at least one key field in the product information.

4. The method according to claim 3, wherein the search server calculates a relevance score for the related product information, in particular:

and the search server calculates the relevance score of the related product information according to a pre-trained relevance ranking model corresponding to the search channel.

5. The method according to claim 2, wherein when the query information is clicked category information, the search server queries related product information related to the query information in the product information according to the query information, specifically: and the search server inquires related product information matched with the category information in all the product information according to the clicked category information.

6. The method according to claim 5, wherein the search server calculates a relevance score for the related product information, in particular:

the search server calculates a relevance score according to a degree of reliability of the product information and the category, the degree of reliability representing a degree of proximity of the product information description to the category.

7. The method of claim 1, wherein the search server calculating a product information feature score for the related product information comprises:

and the search server inquires whether the product information characteristics of the related product information are stored or not, if so, directly calculates the product information characteristic score according to a pre-trained product characteristic sorting model corresponding to the search channel, otherwise, extracts the product information characteristics of each piece of related product information, and calculates the extracted product information characteristic score according to the pre-trained product characteristic sorting model corresponding to the search channel.

8. The method according to claim 7, wherein after extracting the product feature of each piece of related product information, the method further comprises:

and the search server stores the product information characteristics of each piece of extracted relevant product information.

9. A search ranking device, which is applied to a search server of an online trading platform, comprises:

the receiving module is used for receiving query information and search channel information of a user, wherein the query information is query keywords or clicked category information;

the query module is used for querying related product information related to the query information in the search channel;

the calculation module is used for respectively calculating the correlation score and the product information characteristic score of the related product information; the relevance score represents the matching degree of the query information and the product information;

and the sorting module is used for sorting the related product information according to the relevance score and the product information characteristic score.

10. The apparatus of claim 9, wherein the query module comprises:

the information acquisition module is used for acquiring all product information of the search channel according to the search channel information;

and the information query module is used for querying the related product information related to the query information in all the product information according to the query information.

11. The apparatus of claim 10, wherein when the query information is a query keyword, the information query module comprises:

the field acquisition module is used for acquiring at least one key field according to the query keyword;

and the product query module is used for querying the related product information matched with the at least one key field in all the product information.

12. The apparatus according to claim 11, wherein the calculation module is specifically configured to calculate the relevance score for the related product information according to a pre-trained ranking model corresponding to a search channel; and the number of the first and second groups,

and the product characteristic score is calculated according to the relevant product information and a pre-trained product characteristic sorting model corresponding to the search channel.

13. The apparatus according to claim 10, wherein when the query information is clicked category information, the information query module is specifically configured to query, according to the clicked category information of the user, relevant product information related to the category information from all the product information.

14. The apparatus of claim 13, wherein the computing module is specifically configured to compute the relevance score based on a degree of reliability of the product information with respect to the category, the degree of reliability indicating a proximity of the product information description to the category; and the number of the first and second groups,

and the product characteristic score is calculated according to the relevant product information and a pre-trained product characteristic sorting model corresponding to the search channel.

15. The apparatus according to claim 12 or 14, wherein the computing module comprises:

the query judging module is used for querying whether the product characteristics of the related product information are stored or not;

the characteristic extraction module is used for extracting the product characteristics of the related product information when the result of the query judgment module is negative;

the characteristic calculation module is used for directly calculating the characteristic score of the product information according to a pre-trained product characteristic sorting model corresponding to the search channel when the result of the query judgment module is positive; or, when the result of the query judgment module is negative, calculating the product information feature score extracted by the feature extraction module according to a pre-trained product feature sorting model corresponding to the search channel.

16. The apparatus of claim 15, further comprising:

and the characteristic storage module is used for storing the product information characteristics of each piece of related product information extracted by the characteristic extraction module.