CN113868443B - A multimedia resource recommendation method, device and storage medium - Google Patents
A multimedia resource recommendation method, device and storage medium Download PDFInfo
- Publication number
- CN113868443B CN113868443B CN202010623962.8A CN202010623962A CN113868443B CN 113868443 B CN113868443 B CN 113868443B CN 202010623962 A CN202010623962 A CN 202010623962A CN 113868443 B CN113868443 B CN 113868443B
- Authority
- CN
- China
- Prior art keywords
- multimedia
- candidate
- multimedia resources
- resources
- multimedia resource
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/40—Information retrieval; Database structures therefor; File system structures therefor of multimedia data, e.g. slideshows comprising image and additional audio data
- G06F16/43—Querying
- G06F16/435—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/40—Information retrieval; Database structures therefor; File system structures therefor of multimedia data, e.g. slideshows comprising image and additional audio data
- G06F16/43—Querying
- G06F16/438—Presentation of query results
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
本公开公开了一种多媒体资源推荐方法、装置及存储介质，涉及移动互联网技术领域，以至少解决相关技术中，推荐多媒体资源的准确性低的问题。该方法中，根据候选多媒体资源的特征信息，通过多维度选取多媒体资源，得到不同维度下的多媒体资源集合，并将各维度选取的多媒体资源进行综合排序，得到最终的筛选结果。这样，通过多维度选取多媒体资源，可以保证优质的候选多媒体资源不被过滤掉，从而提高了推荐多媒体资源的准确性。
The present disclosure discloses a multimedia resource recommendation method, device and storage medium, which relates to the field of mobile Internet technology, in order to at least solve the problem of low accuracy of recommending multimedia resources in related technologies. In this method, multimedia resources are selected through multiple dimensions according to the feature information of candidate multimedia resources, a set of multimedia resources under different dimensions is obtained, and the multimedia resources selected in each dimension are comprehensively sorted to obtain the final screening result. In this way, by selecting multimedia resources in multiple dimensions, it can be ensured that high-quality candidate multimedia resources are not filtered out, thereby improving the accuracy of recommending multimedia resources.
Description
Technical Field
The disclosure relates to the technical field of mobile internet, and in particular relates to a multimedia resource recommendation method, a device and a storage medium.
Background
With the development of the age, electronic devices have become a part of life, and in order to meet the use demands of users, various application software is generally downloaded and installed in the electronic devices. The user may watch video, listen to music, etc. through the application software. In order to recommend multimedia resources such as videos, music or news and the like which are interested in the user to the user, the application software screens the multimedia resources which are interested in the user from a resource library through a recommendation algorithm.
However, the existing recommendation algorithm acquires the candidate multimedia resource set, and then filters the candidate multimedia resources according to a unified rule to determine a recommendation result, for example, filters according to the number of times of displaying the candidate multimedia resources. However, as the candidate multimedia resources are filtered by using a strong rule, a part of high-quality candidate multimedia resources are killed by mistake, and experiments find that the proportion of false kills is higher, the accuracy of recommending the multimedia resources by the existing recommendation method is lower.
Disclosure of Invention
The embodiment of the disclosure provides a multimedia resource recommendation method, a device and a storage medium, so as to improve the accuracy of recommending multimedia resources.
According to a first aspect of an embodiment of the present disclosure, there is provided a multimedia resource recommendation method, including:
acquiring a candidate multimedia resource set, wherein the candidate multimedia resource set comprises at least two candidate multimedia resources;
screening the candidate multimedia resource set according to the characteristic information of the candidate multimedia resource in at least two dimensions to obtain a multimedia resource set to be recommended corresponding to each dimension, wherein the characteristic information is used for representing the state information and attribute information of the candidate multimedia resource;
Correcting the weight of the multimedia resources in each multimedia resource set to be recommended by using a preset weight factor to obtain screening parameters of each multimedia resource;
and screening the multimedia resources according to the screening parameters.
In one possible implementation manner, the at least two dimensions include the historical display times of the candidate multimedia resources, and the multimedia resource set to be recommended corresponding to the dimension is obtained by the following manner:
Acquiring historical display times of the candidate multimedia resources from the characteristic information of the candidate multimedia resources;
Classifying the candidate multimedia resources according to preset historical display times and the historical display times of the candidate multimedia resources;
performing Bayesian statistical calculation on the historical display times of the candidate multimedia resources in each class, and determining a first threshold value of each class;
And taking the set of candidate multimedia resources with the historical display times larger than the first threshold value in each class as a first set of multimedia resources to be recommended.
In one possible implementation manner, the at least two dimensions include characteristic parameters of the candidate multimedia resources, and the set of multimedia resources to be recommended corresponding to the dimension is obtained by the following manner:
determining characteristic parameters of the candidate multimedia resources by extracting the characteristics of the characteristic information of the candidate multimedia resources;
and taking the set of candidate multimedia resources with the characteristic parameters larger than a second threshold value as a second set of multimedia resources to be recommended.
In one possible implementation manner, the at least two dimensions include the enhancement parameters of the candidate multimedia resources, and the set of multimedia resources to be recommended corresponding to the dimension is obtained by the following manner:
Acquiring the life cycle state of the candidate multimedia resource from the characteristic information of the candidate multimedia resource;
Inputting the life cycle state of the candidate multimedia resource into a life cycle reinforcement learning neural network model to obtain reinforcement parameters of the candidate multimedia resource;
and taking the candidate multimedia resource set with the strengthening parameter larger than the third threshold value as a third multimedia resource set to be recommended.
In one possible implementation manner, if the same multimedia resources exist in each set of multimedia resources to be recommended, modifying the weight of the multimedia resources in each set of multimedia resources to be recommended by a preset weight factor to obtain screening parameters of each multimedia resource, including:
correcting the weight of the multimedia resources in the multimedia resource set to be recommended corresponding to each dimension screening condition by using a preset weight factor to obtain correction parameters of each multimedia resource;
And summing the correction parameters of the same multimedia resources in each multimedia resource set to be recommended to obtain the screening parameters of the multimedia resources.
According to a second aspect of the embodiments of the present disclosure, there is provided a multimedia resource recommendation apparatus, including:
The system comprises a first acquisition unit, a second acquisition unit and a first processing unit, wherein the first acquisition unit is configured to execute acquisition of a candidate multimedia resource set, and the candidate multimedia resource set comprises at least two candidate multimedia resources;
The first screening unit is configured to perform screening on the candidate multimedia resource set according to the characteristic information of the candidate multimedia resources in at least two dimensions to obtain a multimedia resource set to be recommended corresponding to each dimension, wherein the characteristic information is used for representing the state information and attribute information of the candidate multimedia resources;
the correcting unit is configured to execute the correction of the weight of the multimedia resources in each multimedia resource set to be recommended by a preset weight factor to obtain screening parameters of each multimedia resource;
and a second screening unit configured to perform screening of multimedia resources according to the screening parameters.
In one possible implementation manner, at least two dimensions include the historical display times of the candidate multimedia resources, and the multimedia resource set to be recommended corresponding to the dimension is obtained through the following devices:
an acquisition unit configured to perform acquisition of a history of the number of times of presentation of the candidate multimedia asset from the feature information of the candidate multimedia asset;
The classification unit is configured to perform classification on the candidate multimedia resources according to preset historical display times and the historical display times of the candidate multimedia resources;
A calculation unit configured to perform bayesian statistical calculation on the historical display times of the candidate multimedia resources in each class, and determine a first threshold value of each class;
The first determining unit is configured to perform the collection of candidate multimedia resources with the historical showing times larger than the first threshold value in each class as a first multimedia resource collection to be recommended.
In one possible implementation manner, the at least two dimensions include characteristic parameters of the candidate multimedia resources, and the set of multimedia resources to be recommended corresponding to the dimension is obtained through the following means:
A feature extraction unit configured to perform feature extraction of feature information of the candidate multimedia resource, and determine feature parameters of the candidate multimedia resource;
And a second determining unit configured to perform the set of candidate multimedia resources with the characteristic parameter larger than a second threshold value as a second set of multimedia resources to be recommended.
In one possible implementation manner, the at least two dimensions include the enhancement parameters of the candidate multimedia resources, and the multimedia resource set to be recommended corresponding to the dimension is obtained by the following means:
A second acquisition unit configured to perform acquisition of a life cycle state of the candidate multimedia resource from the feature information of the candidate multimedia resource;
An input unit configured to perform inputting a life cycle state of the candidate multimedia resource into a life cycle reinforcement learning neural network model, resulting in reinforcement parameters of the candidate multimedia resource;
And a third determining unit configured to perform the set of candidate multimedia resources with the reinforcement parameter larger than a third threshold as a third set of multimedia resources to be recommended.
In one possible implementation, if the same multimedia resource exists in each set of multimedia resources to be recommended, the correction unit includes:
The multimedia resource weighting subunit is configured to execute correction of the weight of the multimedia resource in the multimedia resource set to be recommended corresponding to each dimension screening condition by using a preset weight factor to obtain correction parameters of each multimedia resource;
And the summation subunit is configured to perform summation of the correction parameters of the same multimedia resources in each multimedia resource set to be recommended to obtain the screening parameters of the multimedia resources.
According to a third aspect of embodiments of the present disclosure, there is provided an electronic device, comprising:
A processor;
a memory for storing the processor-executable instructions;
wherein the processor is configured to execute the instructions to implement a multimedia resource recommendation method;
According to a fourth aspect of embodiments of the present disclosure, there is provided a storage medium, which when executed by a processor of an electronic device, enables the electronic device to perform a multimedia asset recommendation method;
According to a fifth aspect of the disclosed embodiments, there is provided a computer program product comprising at least one processor, and a memory communicatively coupled to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the multimedia resource recommendation method provided by the disclosed embodiments.
The technical scheme provided by the embodiment of the disclosure at least brings the following beneficial effects:
According to the characteristic information of the candidate multimedia resources, selecting the multimedia resources through multiple dimensions to obtain a multimedia resource set under different dimensions, and comprehensively sequencing the multimedia resources selected in each dimension to obtain a final screening result. Therefore, the multimedia resources are selected through multiple dimensions, so that the candidate multimedia resources with high quality can be ensured not to be filtered, and the accuracy of recommending the multimedia resources is improved.
Additional features and advantages of the disclosure will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the disclosure. The objectives and other advantages of the disclosure will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.
Drawings
The accompanying drawings, which are included to provide a further understanding of the disclosure, illustrate and explain the present disclosure, and together with the description serve to explain the present disclosure. In the drawings:
FIG. 1 is a flow chart of a recommendation method in the prior art;
fig. 2 is a flow chart of a multimedia resource recommendation method in an embodiment of the disclosure;
FIG. 3 is a flow chart of a first screening method according to an embodiment of the disclosure;
FIG. 4 is a flow chart of a second screening method according to an embodiment of the disclosure;
FIG. 5 is a flow chart of a third screening method according to an embodiment of the present disclosure;
Fig. 6 is a schematic structural diagram of a multimedia resource recommendation device according to an embodiment of the disclosure;
fig. 7 is a schematic structural diagram of a terminal device in an embodiment of the disclosure.
Detailed Description
In order to improve accuracy of recommending multimedia resources to users, the embodiment of the disclosure provides a multimedia resource recommending method, a device and a storage medium. In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the technical solutions of the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.
It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the foregoing figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the disclosure described herein may be capable of operation in sequences other than those illustrated or described herein. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present disclosure as detailed in the accompanying claims.
The following describes the technical scheme provided by the embodiments of the present disclosure with reference to the accompanying drawings.
In the information age context, both information consumers and information producers have met with significant challenges. It is a difficult matter how to find information of interest to the information consumer from a large amount of information, and how to make the information of interest stand out from the information producer, and it is also a difficult matter how to receive attention of the vast users. The recommendation system is the main tool for solving the contradiction.
The recommendation system can recommend multimedia information which is liked by the information consumer to the information consumer, and also can recommend high-quality multimedia resources produced by the information producer to the information consumer. Taking short videos as an example, in a mature video community, a producer can continuously upload videos shot by himself, a recommendation system needs to filter out works which are low in quality and do not accord with the main stream of the community under the complex condition of information overload, and high-quality works are reasonably screened out to serve as candidate sets of recommendation models, so that the works which are pushed to users to like are further realized, and personalized interests and diversity experience of the users are simultaneously met.
An industrial recommendation system mainly comprises two stages of recall and sorting, wherein a screening process exists between the two stages, and multimedia resources in a recall source are screened and sorted.
The existing recommendation algorithm obtains a candidate multimedia resource set from a recall source, filters the candidate multimedia resource according to a unified rule, determines a recommendation result, and recommends a user after sequencing the recommendation result. As shown in fig. 1, which is a recommended method in the prior art. And after receiving the recommendation request, screening the multimedia resources from the i2i recall source, the u2i recall source and the FM recall source according to a unified rule, and sequencing the screened multimedia resources to obtain a recommendation result. However, as the candidate multimedia resources are filtered by using a strong rule, a part of high-quality candidate multimedia resources are killed by mistake, and experiments find that the proportion of false kills is higher, the accuracy of recommending the multimedia resources by the existing recommendation method is lower.
In view of this, the present disclosure provides a multimedia resource recommendation method, which selects multimedia resources in multiple dimensions to obtain a multimedia resource set in different dimensions, and comprehensively sorts the multimedia resources selected in each dimension to obtain a final screening result. The multimedia resources are selected through multiple dimensions, so that high-quality candidate multimedia resources can be guaranteed not to be filtered, and accuracy of recommending the multimedia resources is improved.
For easy understanding, the technical scheme provided by the present disclosure is further described below with reference to the accompanying drawings.
Fig. 2 is a flowchart illustrating a multimedia asset recommendation method according to an exemplary embodiment, as shown in fig. 2, including the following steps.
In step S21, a candidate multimedia resource set is acquired, wherein the candidate multimedia resource set comprises at least two candidate multimedia resources.
Wherein, the multimedia resources comprise multimedia information such as music, video, news and the like.
The candidate multimedia resource set is obtained by roughly sequencing the multimedia resources in the recall source.
In step S22, the candidate multimedia resource set is filtered according to the feature information of the candidate multimedia resource in at least two dimensions, so as to obtain a multimedia resource set to be recommended corresponding to each dimension, where the feature information is used to characterize the state information and attribute information of the candidate multimedia resource.
In the embodiment of the disclosure, after the candidate multimedia resource set is acquired, the candidate multimedia resource set is screened in multiple dimensions according to the characteristic information of the candidate multimedia resource. The feature information of the candidate multimedia resource comprises attribute type features and statistical type features, wherein the attribute type features can extract types of different dimensions of the multimedia resource (for example, videos are taken as examples, the types of different dimensions comprise various portrait features of the videos, such as content type features comprising scenery, characters, animals and the like, behavior type features comprising dancing, driving, sports and the like), producer attributes (producer attributes refer to portrait features of video authors, comprising gender, age and region of the video authors) and the like. The statistics type features can extract the features of historical display times, click volume, forwarding volume, praise volume, playing completion degree and the like.
In an embodiment of the present disclosure, the dimensions include any of the following:
The historical display times of the first and candidate multimedia resources;
In the embodiment of the present disclosure, when screening candidate multimedia resources with the historical display times of the candidate multimedia resources as dimensions, the candidate multimedia resources may be classified according to the historical display times of the candidate multimedia resources, and candidate multimedia resources meeting the conditions in each class may be obtained, which may be specifically implemented as steps A1-A4:
And A1, acquiring historical display times of the candidate multimedia resources from the characteristic information of the candidate multimedia resources.
And step A2, classifying the candidate multimedia resources according to the preset historical display times and the historical display times of the candidate multimedia resources.
Wherein, the preset historical display times are at least one. Dividing the competition track according to preset historical display times, wherein the preset historical display times are determined based on a statistical learning method.
For example, if the number of preset history display times is two and is 100 and 1000 respectively, three competing tracks are divided according to the preset history display times, and the range of each track is 0-100, 100-1000 and 1000-infinity respectively. And the candidate multimedia resources are divided into corresponding tracks according to the historical display times of the candidate multimedia resources.
And A3, performing Bayesian statistical calculation on the historical display times of the candidate multimedia resources in each class, and determining a first threshold value of each class.
Wherein each track has a first threshold.
And A4, taking the set of candidate multimedia resources with the historical display times larger than the first threshold value in each class as a first set of multimedia resources to be recommended.
In the embodiment of the disclosure, candidate multimedia resources are screened according to the size of historical display times of comparing the first threshold value with the candidate multimedia resources. For example, the two-way track is divided into three tracks, the range of each track is 0-100, 100-1000 and 1000-infinity, and the first threshold value of each track is 80, 860 and 5000. And selecting the multimedia resources with the history display times larger than 80 from the first track, selecting the multimedia resources with the history display times larger than 860 from the second track, selecting the multimedia resources with the history display times larger than 5000 from the third track, and taking the multimedia resources screened by the three tracks as a first multimedia resource set to be recommended.
As shown in fig. 3, a flow chart of the first screening method is shown. The candidate multimedia resources are obtained by roughly sequencing the multimedia resources in various recall sources.
The method comprises the steps of calculating a first threshold value in each track by utilizing a Bayesian statistical model through developing a plurality of tracks, and screening out multimedia resources to be recommended according to the first threshold value of each track. Therefore, candidate multimedia resources with different display times are screened by using different standards, so that the quality of the multimedia resources to be recommended can be improved, and the accuracy of recommending the multimedia resources is further improved.
Second, characteristic parameters of candidate multimedia resources;
The characteristic parameters are obtained by extracting the characteristics of the characteristic information of the candidate multimedia resources.
In the embodiment of the present disclosure, when screening candidate multimedia resources with the feature parameters of the candidate multimedia resources, screening the multimedia resources to be recommended according to the magnitude relation between the feature parameters of the candidate multimedia resources and the second threshold may be specifically implemented as steps B1-B2:
And step B1, determining characteristic parameters of the candidate multimedia resources by extracting the characteristics of the characteristic information of the candidate multimedia resources.
Feature extraction is carried out on the feature information of the candidate multimedia resources to obtain a feature vector representing the feature information of the candidate multimedia resources, and the feature vector is scored through a scoring formula trained by a supervised learning method to obtain the feature parameters of the candidate multimedia resources.
Along with the increase of platform flow, the continuous evolution and the excavation of user interests lead to obvious fluctuation of historical statistics indexes of the multimedia resources, and meanwhile, the system operates at certain stages of a product period, and is expected to introduce specific characteristic fine content to adjust the sensory adjustability of the multimedia resources. A scoring formula is trained by using the quality marking data of the multimedia resources collected before and a part of characteristic fine data marked manually, and the result of the formula can be used as the characteristic parameter of the multimedia resources. The learning algorithm may employ rulefit (rule fit) which is very interpretative.
And B2, taking the set of candidate multimedia resources with the characteristic parameters larger than a second threshold value as a second set of multimedia resources to be recommended.
As shown in fig. 4, a flow chart of the second screening method is shown. And the candidate multimedia resources are subjected to feature extraction to obtain feature parameters, and the feature parameters are compared with a second threshold value to determine a second multimedia resource set to be recommended.
In this way, the characteristic parameters of the candidate multimedia resources are used as important judging basis for screening the multimedia resources, the generalization performance of the original algorithm is improved, and the long-term effect stability is ensured, so that the quality of the multimedia resources to be recommended is improved, and the accuracy of recommending the multimedia resources is further improved.
Thirdly, strengthening parameters of candidate multimedia resources;
wherein the strengthening parameter is obtained according to the life cycle state of the candidate multimedia resource.
In the embodiment of the present disclosure, when screening candidate multimedia resources with the enhancement parameters of the candidate multimedia resources, screening the multimedia resources to be recommended according to the magnitude relation between the enhancement parameters of the candidate multimedia resources and the third threshold may be specifically implemented as steps C1 to C3:
and step C1, acquiring the life cycle state of the candidate multimedia resource from the characteristic information of the candidate multimedia resource.
As with anything, a multimedia resource also undergoes incubation, birth, growth, maturation, decay, etc., and is generally referred to as a life cycle of the multimedia resource, and it can be determined which stage of the life cycle the candidate multimedia resource is in by the characteristic information of the candidate multimedia resource.
And step C2, inputting the life cycle state of the candidate multimedia resource into a life cycle reinforcement learning neural network model to obtain reinforcement parameters of the candidate multimedia resource.
And step C3, taking the set of candidate multimedia resources with the strengthening parameter larger than a third threshold value as a third set of multimedia resources to be recommended.
As shown in fig. 5, a flow chart of a third screening method is shown. And finally, comparing the strengthening parameter with a third threshold value to determine a third multimedia resource set to be recommended.
In step S23, the weight of the multimedia resources in each set of multimedia resources to be recommended is modified by a preset weight factor, so as to obtain screening parameters of each multimedia resource.
In step S24, the multimedia resources are screened according to the screening parameters.
In the embodiment of the present disclosure, after the first set of multimedia resources to be recommended, the second set of multimedia resources to be recommended, and the third set of multimedia resources to be recommended are obtained, the three sets need to be screened again.
Wherein the preset weight may be preset, or may be empirically determined. The weight of the multimedia resource is treated as 1 before unmodified.
In the embodiment of the disclosure, the preset weight factor is multiplied by the weight of the multimedia resource to obtain the screening parameter of the multimedia resource. For example, if there are 3 multimedia resources, a1, a2 and a3, and their corresponding weight factors are w1, w2 and w3, respectively, then, after correction, the screening parameters of the 3 multimedia resources are a1w1, a2w2 and a3w3, respectively. And screening the multimedia resources according to the obtained screening parameters. The multimedia resources may be selected according to the magnitude of the value of the filtering parameter, for example, the multimedia resources with the top order are selected according to the order of arrangement from big to small. The multimedia resources may also be selected from multimedia resources whose screening parameters exceed a threshold, which is not limiting to the present disclosure.
In the screening, the multimedia resources in all dimensions are screened.
Thus, the quality of the multimedia resources can be further improved by screening again, so that the accuracy of recommending the multimedia resources is improved.
However, in the embodiment of the present disclosure, in the set of multimedia resources to be recommended determined through the three dimensions, there may be one multimedia resource that exists in multiple sets of multimedia resources to be recommended at the same time, and at this time, the multimedia resources need to be weighted and summed to obtain a final filtering parameter, which may be specifically implemented as follows:
correcting the weight of the multimedia resources in the multimedia resource set to be recommended corresponding to each dimension screening condition by using a preset weight factor to obtain correction parameters of each multimedia resource, and summing the correction parameters of the same multimedia resources in each multimedia resource set to be recommended to obtain the screening parameters of the multimedia resources.
For example, if there is one multimedia resource, all of which are in the multimedia resource set to be recommended in three dimensions, the correction parameters in each dimension are determined first, and the correction parameters in each dimension are summed up to obtain the screening parameters. If the preset weight of a multimedia resource a1 in the first dimension is w1, the preset weight in the second dimension is w2, and the preset weight in the third dimension is w3, the filtering parameter of the multimedia resource is (w1+w2+w3) a1.
In this way, the same multimedia resources in different sets are weighted and summed, so that high-quality multimedia resources can be determined more accurately, and the accuracy of recommending the multimedia resources can be improved.
And selecting the multimedia resources through multiple dimensions to obtain a multimedia resource set under different dimensions, and comprehensively sequencing the multimedia resources selected by each dimension to obtain a final screening result. Therefore, the multimedia resources are selected through multiple dimensions, and the high-quality candidate multimedia resources can be ensured not to be filtered, so that the overall quality of the multimedia resources is improved, and the accuracy of recommending the multimedia resources is further improved.
Based on the same inventive concept, the disclosure also provides a multimedia resource recommendation device. Fig. 6 is a schematic diagram of a multimedia resource recommendation device provided in the present disclosure. The device comprises:
a first obtaining unit 601 configured to perform obtaining a candidate multimedia resource set, where the candidate multimedia resource set includes at least two candidate multimedia resources;
A first filtering unit 602, configured to perform filtering on the candidate multimedia resource set in at least two dimensions according to feature information of the candidate multimedia resource, to obtain a multimedia resource set to be recommended corresponding to each dimension, where the feature information is used to characterize state information and attribute information of the candidate multimedia resource;
A correction unit 603 configured to perform correction on the weights of the multimedia resources in each set of multimedia resources to be recommended by using a preset weight factor, so as to obtain screening parameters of each multimedia resource;
a second screening unit 604 configured to perform screening of multimedia resources according to the screening parameters.
In one possible implementation manner, at least two dimensions include the historical display times of the candidate multimedia resources, and the multimedia resource set to be recommended corresponding to the dimension is obtained through the following devices:
an acquisition unit configured to perform acquisition of a history of the number of times of presentation of the candidate multimedia asset from the feature information of the candidate multimedia asset;
The classification unit is configured to perform classification on the candidate multimedia resources according to preset historical display times and the historical display times of the candidate multimedia resources;
A calculation unit configured to perform bayesian statistical calculation on the historical display times of the candidate multimedia resources in each class, and determine a first threshold value of each class;
The first determining unit is configured to perform the collection of candidate multimedia resources with the historical showing times larger than the first threshold value in each class as a first multimedia resource collection to be recommended.
In one possible implementation manner, the at least two dimensions include characteristic parameters of the candidate multimedia resources, and the set of multimedia resources to be recommended corresponding to the dimension is obtained through the following means:
A feature extraction unit configured to perform feature extraction of feature information of the candidate multimedia resource, and determine feature parameters of the candidate multimedia resource;
And a second determining unit configured to perform the set of candidate multimedia resources with the characteristic parameter larger than a second threshold value as a second set of multimedia resources to be recommended.
In one possible implementation manner, the at least two dimensions include the enhancement parameters of the candidate multimedia resources, and the multimedia resource set to be recommended corresponding to the dimension is obtained by the following means:
A second acquisition unit configured to perform acquisition of a life cycle state of the candidate multimedia resource from the feature information of the candidate multimedia resource;
An input unit configured to perform inputting a life cycle state of the candidate multimedia resource into a life cycle reinforcement learning neural network model, resulting in reinforcement parameters of the candidate multimedia resource;
And a third determining unit configured to perform the set of candidate multimedia resources with the reinforcement parameter larger than a third threshold as a third set of multimedia resources to be recommended.
In one possible implementation, if the same multimedia resource exists in each set of multimedia resources to be recommended, the modifying unit 603 includes:
The multimedia resource weighting subunit is configured to execute correction of the weight of the multimedia resource in the multimedia resource set to be recommended corresponding to each dimension screening condition by using a preset weight factor to obtain correction parameters of each multimedia resource;
And the summation subunit is configured to perform summation of the correction parameters of the same multimedia resources in each multimedia resource set to be recommended to obtain the screening parameters of the multimedia resources.
As shown in fig. 7, the embodiment of the present disclosure further provides an electronic device 70, which may include a memory 701 and a processor 702, based on the same technical concept.
The memory 701 is used for storing a computer program executed by the processor 702. The memory 701 may mainly include a storage program area which may store an operating system, an application program required for at least one function, and the like, and a storage data area which may store data created according to the use of the task management device, and the like. The processor 702 may be a central processing unit (central processing unit, CPU), or a digital processing unit, or the like. The particular connection medium between the memory 701 and the processor 702 described above is not limited in the presently disclosed embodiments. The embodiment of the present disclosure is shown in fig. 7, where the memory 701 and the processor 702 are connected by a bus 703, where the bus 703 is shown in bold lines in fig. 7, and the connection between other components is merely illustrative, and not limiting. The bus 703 may be classified into an address bus, a data bus, a control bus, and the like. For ease of illustration, only one thick line is shown in fig. 7, but not only one bus or one type of bus.
The memory 701 may be a volatile memory (RAM) such as a random-access memory (RAM), the memory 701 may be a nonvolatile memory (non-volatile memory) such as a read-only memory (rom), a flash memory (flash memory), a hard disk (HARD DISK DRIVE, HDD) or a solid state disk (solid-state drive) (STATE DRIVE, SSD), or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited thereto. Memory 701 may be a combination of the above.
A processor 702 for executing the method performed by the device in the embodiment shown in fig. 2 when invoking the computer program stored in said memory 701.
In some possible implementations, aspects of the methods provided by the present disclosure may also be implemented in the form of a program product comprising program code for causing a computer device to carry out the steps of the methods according to the various exemplary embodiments of the disclosure described above when the program product is run on the computer device, e.g. the computer device may carry out the methods as carried out by the devices in the examples shown in fig. 2-6.
The program product may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of a readable storage medium include an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
While the preferred embodiments of the present disclosure have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the disclosure. Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This disclosure is intended to cover any adaptations, uses, or adaptations of the disclosure following the general principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
It is to be understood that the present disclosure is not limited to the precise arrangements and instrumentalities shown in the drawings, and that various modifications and changes may be effected without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.
Claims (10)
1. A method for recommending multimedia resources, the method comprising:
Acquiring a candidate multimedia resource set, wherein the candidate multimedia resource set comprises at least three candidate multimedia resources;
screening the candidate multimedia resource set according to the characteristic information of the candidate multimedia resources by at least three dimensions to obtain a multimedia resource set to be recommended corresponding to each dimension, wherein the characteristic information is used for representing the state information and attribute information of the candidate multimedia resources;
Correcting the weight of the multimedia resources in each multimedia resource set to be recommended by using a preset weight factor to obtain screening parameters of each multimedia resource;
Screening multimedia resources according to the screening parameters;
the at least three dimensions comprise a historical display times dimension of the candidate multimedia resource, a characteristic parameter dimension of the candidate multimedia resource and a strengthening parameter dimension of the candidate multimedia resource;
The feature parameters in the feature parameter dimension of the candidate multimedia resource are obtained by extracting features of feature information of the candidate multimedia resource to obtain feature vectors used for representing the feature information, and scoring the feature vectors through a scoring formula trained by a supervised learning method;
if the same multimedia resources exist in each multimedia resource set to be recommended, correcting the weight of the multimedia resources in each multimedia resource set to be recommended by a preset weight factor to obtain screening parameters of the multimedia resources, wherein the screening parameters comprise:
correcting the weight of the multimedia resources in the multimedia resource set to be recommended corresponding to each dimension screening condition by using a preset weight factor to obtain correction parameters of each multimedia resource;
And summing the correction parameters of the same multimedia resources in each multimedia resource set to be recommended to obtain the screening parameters of the multimedia resources.
2. The method for recommending multimedia resources according to claim 1, wherein the set of multimedia resources to be recommended corresponding to the historical display times dimension of the candidate multimedia resources is obtained by:
Acquiring historical display times of the candidate multimedia resources from the characteristic information of the candidate multimedia resources;
Classifying the candidate multimedia resources according to preset historical display times and the historical display times of the candidate multimedia resources;
performing Bayesian statistical calculation on the historical display times of the candidate multimedia resources in each class, and determining a first threshold value of each class;
And taking the set of candidate multimedia resources with the historical display times larger than the first threshold value in each class as a first set of multimedia resources to be recommended.
3. The method for recommending multimedia resources according to claim 1, wherein the set of multimedia resources to be recommended corresponding to the feature parameter dimension of the candidate multimedia resources is obtained by:
And taking the set of candidate multimedia resources with the characteristic parameters larger than a second threshold value as a second set of multimedia resources to be recommended.
4. The method for recommending multimedia resources according to claim 1, wherein the set of multimedia resources to be recommended corresponding to the reinforcement parameter dimension of the candidate multimedia resources is obtained by:
Acquiring the life cycle state of the candidate multimedia resource from the characteristic information of the candidate multimedia resource;
Inputting the life cycle state of the candidate multimedia resource into a life cycle reinforcement learning neural network model to obtain reinforcement parameters of the candidate multimedia resource;
and taking the candidate multimedia resource set with the strengthening parameter larger than the third threshold value as a third multimedia resource set to be recommended.
5. A multimedia asset recommendation device, comprising:
the system comprises a first acquisition unit, a second acquisition unit and a first processing unit, wherein the first acquisition unit is configured to execute acquisition of a candidate multimedia resource set, and the candidate multimedia resource set comprises at least three candidate multimedia resources;
The first screening unit is configured to perform screening on the candidate multimedia resource set according to the characteristic information of the candidate multimedia resources in at least three dimensions to obtain a multimedia resource set to be recommended corresponding to each dimension, wherein the characteristic information is used for representing the state information and attribute information of the candidate multimedia resources;
the correcting unit is configured to execute the correction of the weight of the multimedia resources in each multimedia resource set to be recommended by a preset weight factor to obtain screening parameters of each multimedia resource;
a second screening unit configured to perform screening of multimedia resources according to the screening parameters;
the at least three dimensions comprise a historical display times dimension of the candidate multimedia resource, a characteristic parameter dimension of the candidate multimedia resource and a strengthening parameter dimension of the candidate multimedia resource;
The feature parameters in the feature parameter dimension of the candidate multimedia resource are obtained by extracting features of feature information of the candidate multimedia resource to obtain feature vectors used for representing the feature information, and scoring the feature vectors through a scoring formula trained by a supervised learning method;
if the same multimedia resources exist in each multimedia resource set to be recommended, the correction unit includes:
The multimedia resource weighting subunit is configured to execute correction of the weight of the multimedia resource in the multimedia resource set to be recommended corresponding to each dimension screening condition by using a preset weight factor to obtain correction parameters of each multimedia resource;
And the summation subunit is configured to perform summation of the correction parameters of the same multimedia resources in each multimedia resource set to be recommended to obtain the screening parameters of the multimedia resources.
6. The apparatus of claim 5, wherein the set of multimedia resources to be recommended corresponding to the historical display times dimension of the candidate multimedia resources is obtained by:
an acquisition unit configured to perform acquisition of a history of the number of times of presentation of the candidate multimedia asset from the feature information of the candidate multimedia asset;
The classification unit is configured to perform classification on the candidate multimedia resources according to preset historical display times and the historical display times of the candidate multimedia resources;
A calculation unit configured to perform bayesian statistical calculation on the historical display times of the candidate multimedia resources in each class, and determine a first threshold value of each class;
The first determining unit is configured to perform the collection of candidate multimedia resources with the historical showing times larger than the first threshold value in each class as a first multimedia resource collection to be recommended.
7. The apparatus of claim 5, wherein the set of multimedia resources to be recommended corresponding to the feature parameter dimension of the candidate multimedia resources is obtained by:
And the second determining unit is configured to perform the collection of the candidate multimedia resources with the characteristic parameters of the candidate multimedia resources larger than a second threshold value as a second collection of the multimedia resources to be recommended.
8. The apparatus of claim 5, wherein the set of multimedia resources to be recommended corresponding to the enhanced parameter dimension of the candidate multimedia resources is obtained by:
A second acquisition unit configured to perform acquisition of a life cycle state of the candidate multimedia resource from the feature information of the candidate multimedia resource;
An input unit configured to perform inputting a life cycle state of the candidate multimedia resource into a life cycle reinforcement learning neural network model, resulting in reinforcement parameters of the candidate multimedia resource;
And a third determining unit configured to perform the set of candidate multimedia resources with the reinforcement parameter larger than a third threshold as a third set of multimedia resources to be recommended.
9. An electronic device, comprising:
A processor;
a memory for storing the processor-executable instructions;
wherein the processor is configured to execute the instructions to implement the multimedia asset recommendation method of any of claims 1 to 4.
10. A storage medium, wherein instructions in the storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the multimedia asset recommendation method of any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010623962.8A CN113868443B (en) | 2020-06-30 | 2020-06-30 | A multimedia resource recommendation method, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010623962.8A CN113868443B (en) | 2020-06-30 | 2020-06-30 | A multimedia resource recommendation method, device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113868443A CN113868443A (en) | 2021-12-31 |
| CN113868443B true CN113868443B (en) | 2024-12-20 |
Family
ID=78981023
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010623962.8A Active CN113868443B (en) | 2020-06-30 | 2020-06-30 | A multimedia resource recommendation method, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113868443B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114595406A (en) * | 2022-03-17 | 2022-06-07 | 阿里巴巴（中国）有限公司 | Sorting method and device for multiple recall results |
| CN117009644A (en) * | 2022-10-27 | 2023-11-07 | 腾讯科技（深圳）有限公司 | Multimedia information recommendation method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108228824A (en) * | 2017-12-29 | 2018-06-29 | 暴风集团股份有限公司 | Recommendation method, apparatus, electronic equipment, medium and the program of a kind of video |
| CN109639786A (en) * | 2018-12-04 | 2019-04-16 | 北京达佳互联信息技术有限公司 | Distribution method, device, server and the storage medium of multimedia resource |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8849095B2 (en) * | 2011-07-26 | 2014-09-30 | Ooyala, Inc. | Goal-based video delivery system |
| CN102663010A (en) * | 2012-03-20 | 2012-09-12 | 复旦大学 | Personalized image browsing and recommending method based on labelling semantics and system thereof |
| CN106168980B (en) * | 2016-07-26 | 2020-07-28 | 阿里巴巴（中国）有限公司 | Multimedia resource recommendation sequencing method and device |
| CN108073303B (en) * | 2016-11-17 | 2021-11-30 | 北京搜狗科技发展有限公司 | Input method and device and electronic equipment |
| US10368132B2 (en) * | 2016-11-30 | 2019-07-30 | Facebook, Inc. | Recommendation system to enhance video content recommendation |
| CN106802915B (en) * | 2016-12-09 | 2020-07-28 | 宁波大学 | Academic resource recommendation method based on user behaviors |
| WO2019000472A1 (en) * | 2017-06-30 | 2019-01-03 | 广东欧珀移动通信有限公司 | Navigation method and apparatus, storage medium, and server |
| CN108665148B (en) * | 2018-04-18 | 2022-02-22 | 腾讯科技（深圳）有限公司 | Electronic resource quality evaluation method and device and storage medium |
| CN110929052B (en) * | 2019-12-03 | 2023-04-18 | 北京奇艺世纪科技有限公司 | Multimedia resource recommendation method and device, electronic equipment and storage medium |
| CN111143697B (en) * | 2020-01-02 | 2023-03-21 | 腾讯科技（深圳）有限公司 | Content recommendation method and related device |
-
2020
- 2020-06-30 CN CN202010623962.8A patent/CN113868443B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108228824A (en) * | 2017-12-29 | 2018-06-29 | 暴风集团股份有限公司 | Recommendation method, apparatus, electronic equipment, medium and the program of a kind of video |
| CN109639786A (en) * | 2018-12-04 | 2019-04-16 | 北京达佳互联信息技术有限公司 | Distribution method, device, server and the storage medium of multimedia resource |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113868443A (en) | 2021-12-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110704674B (en) | Video playing integrity prediction method and device | |
| CN110781321B (en) | Multimedia content recommendation method and device | |
| CN110442790A (en) | Recommend method, apparatus, server and the storage medium of multi-medium data | |
| CN102262647B (en) | Signal conditioning package, information processing method and program | |
| CN111898675B (en) | Credit wind control model generation method and device, scoring card generation method, machine readable medium and equipment | |
| CN111859149A (en) | Information recommendation method, device, electronic device and storage medium | |
| CN110019943B (en) | Video recommendation method and device, electronic equipment and storage medium | |
| CN112131456B (en) | Information push method, device, equipment and storage medium | |
| CN108073902A (en) | Video summary method, apparatus and terminal device based on deep learning | |
| CN106339507A (en) | Method and device for pushing streaming media message | |
| CN111737473B (en) | Text classification method, device and equipment | |
| CN108595526B (en) | Resource recommendation method and device | |
| CN111816170A (en) | Training of audio classification model and junk audio recognition method and device | |
| CN111177564B (en) | Product recommendation method and device | |
| CN113868443B (en) | A multimedia resource recommendation method, device and storage medium | |
| CN111178656A (en) | Credit model training method, credit scoring device and electronic equipment | |
| CN115599986B (en) | Method, device, intelligent device and storage medium for processing user recommendations | |
| CN106599047A (en) | Information pushing method and device | |
| CN112163163B (en) | Multi-algorithm fused information recommendation method, device and equipment | |
| CN113065067A (en) | Article recommendation method and device, computer equipment and storage medium | |
| CN112801053B (en) | Video data processing method and device | |
| CN114329055A (en) | Search recommendation method and recommendation device, electronic device and storage medium | |
| CN111353052B (en) | Multimedia object recommendation method and device, electronic equipment and storage medium | |
| CN114443943A (en) | Information scheduling method, device and equipment and computer readable storage medium | |
| CN112822527B (en) | Video recommendation method and device, server and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |