An information classification method, device, apparatus and storage medium

Abstract

The present disclosure relates to an information classification method, device, equipment and storage medium, the method comprising: obtaining a search word corresponding to each search operation of a user and behavior data of the user on a search result of the search operation; constructing a first bipartite graph between the search word and multimedia data and a second bipartite graph between the search word and a user account according to the search word and the behavior data; performing feature extraction on the structure of the second bipartite graph to obtain initial features of the search word and account features of the user account; performing feature extraction on the first bipartite graph based on the initial features to obtain updated features of the search word and data features of the multimedia data; and clustering the multimedia data, the user account and the search word according to the account features, the updated features and the data features to obtain a plurality of clusters as a classification result of information. In this way, the number of prior labels required for classification is reduced, and the information of the multimedia data and the user account is effectively coordinated and promoted.

Description

Technical Field

[0001] The present application relates to the field of data classification, and in particular to an information classification method, apparatus, device and storage medium. Background Art

[0002] A large amount of multimedia data is uploaded to the multimedia data platform every day. Among these multimedia data, some is useful multimedia data. Useful multimedia data refers to multimedia data that is helpful to users, such as multimedia data that explains how to make fried rice with egg, how to learn to drive, etc. User accounts that produce more useful multimedia data can be called useful user accounts.

[0003] In some scenarios, users need to search for useful multimedia data or user accounts on a multimedia data platform. Therefore, the multimedia data platform needs to classify the multimedia data and user accounts to identify the useful multimedia data and user accounts. Existing technologies typically use traditional classification methods to identify useful multimedia data and user accounts. This requires labeling large amounts of multimedia data and user accounts, and requires two unrelated classification networks: useful multimedia data classification and useful user account classification.

[0004] Since the classification network relies on prior labeling of a large number of videos and user accounts, it is not accurate enough and cannot effectively coordinate and promote the information of useful multimedia data and useful user accounts. Therefore, the existing technology solutions have low efficiency in classifying multimedia data and user accounts, and the accuracy is difficult to guarantee, which cannot meet user needs. Summary of the Invention

[0005] In order to solve the problems in related technologies of low efficiency and poor accuracy in classifying multimedia data and user accounts, which cannot meet user needs, the present disclosure provides an information classification method, apparatus, device, and storage medium. The technical solutions of the present disclosure are as follows:

[0006] According to a first aspect of an embodiment of the present disclosure, there is provided an information classification method, the method comprising:

[0007] Obtaining behavioral data of a search term and search results for the search term, wherein the search results include multimedia data and a user account;

[0008] constructing, based on the search term and the behavior data, a first bipartite graph between the search term and the multimedia data and a second bipartite graph between the search term and the user account;

[0009] performing feature extraction on the second bipartite graph structure to obtain account features of the user account;

[0010] Performing feature extraction on the first bipartite graph to obtain search features of the search term and data features of the multimedia data;

[0011] The multimedia data, the user account and the search term are clustered according to the account feature, the search feature and the data feature to obtain a plurality of clusters, each cluster including at least one of the multimedia data, the search term or the user account.

[0012] Optionally, after clustering the multimedia data, the user account, and the search term according to the account feature, the search feature, and the data feature to obtain a plurality of clusters, the method further includes:

[0013] A cluster including the multimedia data, the search terms or the user accounts meeting preset conditions is determined as a target cluster, the multimedia data in the target cluster is the target multimedia data, and the user accounts in the target cluster are the target user accounts.

[0014] Optionally, the preset conditions include any one or more of the following:

[0015] The search terms in the cluster are all preset target search terms or contain preset fields;

[0016] The preset target user accounts in the cluster account for more than half of the total number of user accounts in the cluster;

[0017] The amount of the preset target multimedia data in the cluster accounts for more than half of the total amount of multimedia data in the cluster, or the amount of the preset target multimedia data in the cluster is greater than a preset amount threshold.

[0018] Optionally, constructing a first bipartite graph between the search term and the multimedia data and a second bipartite graph between the search term and the user account based on the search term and the behavior data includes:

[0019] Calculating, based on the behavior data, a click-through rate and a satisfaction rate between the search term corresponding to the behavior data and the multimedia data;

[0020] Calculating the relevance between the search term and the multimedia data according to the click-through rate and the satisfaction level;

[0021] When the correlation degree is higher than a preset correlation threshold, a connection is established between the search term and the multimedia data to obtain a first bipartite graph, and a connection is established between the search term and the user account corresponding to the multimedia data to obtain a second bipartite graph.

[0022] Optionally, the behavior data includes the number of times the user displays, clicks, plays, likes, follows, the number of times the play time exceeds a preset time, and play order parameters of the multimedia data;

[0023] Calculating the click-through rate and satisfaction between the search term corresponding to the behavior data and the multimedia data based on the behavior data includes:

[0024] Calculating a ratio between the number of clicks and the number of impressions as a click-through rate between the search term corresponding to the behavior data and the multimedia data;

[0025] Calculating the sum of the number of times the playback duration exceeds a preset duration, the number of likes, the number of follows, and the playback order parameter, and using the ratio of the sum to the number of playbacks as the satisfaction between the search term corresponding to the behavior data and the multimedia data;

[0026] Calculating the correlation between the search term and the multimedia data based on the click-through rate and the satisfaction level includes:

[0027] The product of the click-through rate and the satisfaction level is calculated as the correlation between the search term and the multimedia data.

[0028] Optionally, the performing feature extraction on the second bipartite graph structure to obtain account features of the user account includes:

[0029] Using the first graph neural network, extracting features from the second bipartite graph structure to obtain initial features of the search term and account features of the user account;

[0030] The extracting features of the first bipartite graph to obtain the search features of the search term and the data features of the multimedia data includes:

[0031] A second graph neural network is used to perform feature extraction on the first bipartite graph structure to obtain data features of the multimedia data and search features of the search terms, wherein the second graph neural network is obtained based on the training of the first graph neural network, and the learning rate of the second graph neural network is less than the learning rate of the first graph neural network.

[0032] According to a second aspect of an embodiment of the present disclosure, there is provided an information classification device, the device comprising:

[0033] an acquiring unit configured to acquire behavioral data of a search term and a search result for the search term, wherein the search result includes multimedia data and a user account;

[0034] a construction unit configured to construct, based on the search term and the behavior data, a first bipartite graph between the search term and the multimedia data and a second bipartite graph between the search term and the user account;

[0035] a first extraction unit configured to perform feature extraction on the second bipartite graph structure to obtain initial features of the search term and account features of the user account;

[0036] a second extraction unit configured to perform feature extraction on the first bipartite graph to obtain search features of the search term and data features of the multimedia data;

[0037] The classification unit is configured to perform clustering of the multimedia data, the user account and the search term based on the account characteristics, the search characteristics and the data characteristics to obtain multiple clusters, each cluster including at least one of the multimedia data, the search term or the user account.

[0038] Optionally, the classification unit is configured to perform:

[0039] A cluster including the multimedia data, the search terms or the user accounts meeting preset conditions is determined as a target cluster, the multimedia data in the target cluster is the target multimedia data, and the user accounts in the target cluster are the target user accounts.

[0040] Optionally, the preset conditions include any one or more of the following:

[0041] The search terms in the cluster are all preset target search terms or contain preset fields;

[0042] The preset target user accounts in the cluster account for more than half of the total number of user accounts in the cluster;

[0043] The amount of the preset target multimedia data in the cluster accounts for more than half of the total amount of multimedia data in the cluster, or the amount of the preset target multimedia data in the cluster is greater than a preset amount threshold.

[0044] Optionally, the construction unit is configured to execute:

[0045] Calculating, based on the behavior data, a click-through rate and a satisfaction rate between the search term corresponding to the behavior data and the multimedia data;

[0046] Calculating the relevance between the search term and the multimedia data according to the click-through rate and the satisfaction level;

[0047] When the correlation degree is higher than a preset correlation threshold, a connection is established between the search term and the multimedia data to obtain a first bipartite graph, and a connection is established between the search term and the user account corresponding to the multimedia data to obtain a second bipartite graph.

[0048] Optionally, the behavior data includes the number of times the user displays, clicks, plays, likes, follows, the number of times the play time exceeds a preset time, and play order parameters of the multimedia data;

[0049] The construction unit is configured to execute:

[0050] Calculating a ratio between the number of clicks and the number of impressions as a click-through rate between the search term corresponding to the behavior data and the multimedia data;

[0051] Calculating the sum of the number of times the playback duration exceeds a preset duration, the number of likes, the number of follows, and the playback order parameter, and using the ratio of the sum to the number of playbacks as the satisfaction between the search term corresponding to the behavior data and the multimedia data;

[0052] The product of the click-through rate and the satisfaction level is calculated as the correlation between the search term and the multimedia data.

[0053] Optionally, the first extraction unit is configured to execute:

[0054] Using the first graph neural network, extracting features from the second bipartite graph structure to obtain initial features of the search term and account features of the user account;

[0055] The second extraction unit is configured to execute:

[0056] A second graph neural network is used to perform feature extraction on the first bipartite graph structure to obtain data features of the multimedia data and search features of the search terms, wherein the second graph neural network is obtained based on the training of the first graph neural network, and the learning rate of the second graph neural network is less than the learning rate of the first graph neural network.

[0057] According to a third aspect of the embodiments of the present disclosure, there is further provided an electronic device, including:

[0058] processor;

[0059] a memory for storing instructions executable by the processor;

[0060] The processor is configured to execute the instructions to implement the method described in the first aspect.

[0061] According to a fourth aspect of an embodiment of the present disclosure, a storage medium is further provided. When instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method described in the first aspect.

[0062] According to a fifth aspect of an embodiment of the present disclosure, a computer program product comprising instructions is provided, which, when executed on a computer, enables the computer to implement the method described in the first aspect.

[0063] In the technical solution provided by the embodiments of the present disclosure, behavioral data of search terms and search results for the search terms are obtained, and the search results include multimedia data and user accounts; based on the search terms and behavioral data, a first bipartite graph between the search terms and the multimedia data and a second bipartite graph between the search terms and the user accounts are constructed; feature extraction is performed on the second bipartite graph structure to obtain account features of the user account; feature extraction is performed on the first bipartite graph to obtain search features of the search terms and data features of the multimedia data; based on the account features, search features and data features, the multimedia data, user accounts and search terms are clustered to obtain multiple clusters, and each cluster includes at least one multimedia data, search term or user account.

[0064] In this way, by constructing the first bipartite graph between the search terms and the multimedia data and the second bipartite graph between the search terms and the user accounts, and extracting the features of the multimedia data and the user accounts based on the first bipartite graph and the second bipartite graph, the multimedia data and the user accounts are clustered. On the one hand, the number of prior labels required for classification is reduced and the accuracy of classification is improved. On the other hand, in the process of extracting the search term features, the information of the multimedia data and the information of the user accounts are simultaneously utilized, so that the information of the multimedia data and the user accounts are effectively coordinated and promoted. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 is a flow chart showing an information classification method according to an exemplary embodiment;

[0066] Figure 2 is a block diagram of an information classification device according to an exemplary embodiment;

[0067] Figure 3 is a schematic structural diagram of an electronic device according to an exemplary embodiment;

[0068] Figure 4 It is a block diagram of a device according to an exemplary embodiment. DETAILED DESCRIPTION

[0069] In order to enable ordinary persons in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0070] It should be noted that the terms "first," "second," and the like in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the numbers used in this manner are interchangeable where appropriate so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure as detailed in the appended claims.

[0071] In order to solve the problem in the related art that a server may receive a large number of information acquisition requests at the same time, resulting in excessive load pressure, the embodiments of the present disclosure provide an information classification system, method, apparatus, device and storage medium.

[0072] First, an information classification method provided by an embodiment of the present disclosure will be introduced in detail.

[0073] like Figure 1 As shown, a flowchart of an information classification method provided by an embodiment of the present disclosure specifically includes the following steps.

[0074] In S11 , behavioral data of a search term and search results for the search term are obtained, where the search results include multimedia data and a user account.

[0075] In this disclosure, multimedia data can be video data, image data, or audio data, without specific limitations. Each time a user searches, they enter a search term and obtain corresponding search results. The user can perform actions such as clicking, displaying, playing, liking, and following multimedia data or user accounts in the search results. These actions are recorded to obtain corresponding behavioral data.

[0076] In some scenarios, these multimedia data and user accounts need to be classified. For example, multimedia data can be divided into useful multimedia data and interesting multimedia data, and user accounts can be divided into useful user accounts and interesting user accounts. Based on the classification results, in the subsequent data search and push process, high-quality authors and high-quality videos can be quickly and accurately discovered and mined, thereby enhancing the enthusiasm of high-quality authors in the community. At the same time, users can obtain high-quality resources and create a positive cycle.

[0077] In S12 , a first bipartite graph between the search term and the multimedia data and a second bipartite graph between the search term and the user account are constructed based on the search term and the behavior data.

[0078] In an implementation, constructing the first bipartite graph between the search word and the multimedia data and the second bipartite graph between the search word and the user account requires determining the association between the search word and the multimedia data and the association between the search word and the user account.

[0079] Specifically, the click-through rate and the satisfaction degree between the search word corresponding to the behavior data and the multimedia data can be calculated according to the behavior data; then, the association degree between the search word and the multimedia data is calculated according to the click-through rate and the satisfaction degree; further, the connection between the search word and the multimedia data is established when the association degree is higher than a preset association threshold, and the first bipartite graph is obtained, and the connection between the search word and the user account corresponding to the multimedia data is established, and the second bipartite graph is obtained.

[0080] In this way, the search result that the user is more satisfied with is determined as the multimedia data or the user account corresponding to the search word by analyzing the behavior data, the connection between the search word and the multimedia data or the user account is established, and the bipartite graph is obtained. The subsequent classification result will be more matched with the behavior data of the user, thereby meeting the needs of the user.

[0081] In an implementation, the behavior data includes the display times, the click times, the play times, the like times, the follow times, the times of the play duration exceeding a preset duration, and the play order parameter of the user to the multimedia data.

[0082] Then, according to the behavior data, the click-through rate and the satisfaction degree between the search word corresponding to the behavior data and the multimedia data can include:

[0083] The ratio between the click times and the display times is calculated as the click-through rate between the search word corresponding to the behavior data and the multimedia data; the sum of the times of the play duration exceeding a preset duration, the like times, the follow times, and the play order parameter is calculated, and the ratio between the sum and the play times is calculated as the satisfaction degree between the search word corresponding to the behavior data and the multimedia data.

[0084] According to the click-through rate and the satisfaction degree, the association degree between the search word and the multimedia data can include calculating the product between the click-through rate and the satisfaction degree as the association degree between the search word and the multimedia data.

[0085] For example, the following formula can be used:

[0086] poster=click / show*(long_play+a*like+b*follow+first_click+last_long_play) / play

[0087] Wherein, Show represents the number of displays, click represents the number of clicks, long_play represents the number of times of playing for more than a preset time length, the preset time length is 18 seconds, like represents the number of likes, follow represents the number of attentions, first_click represents a play order parameter corresponding to multimedia data clicked for the first time in one search, last_long_play represents a play order parameter corresponding to multimedia data played for the last time in one search, play represents the number of plays, and poster represents the correlation degree.

[0088] In S13, the second bipartite graph structure is subjected to feature extraction to obtain initial features of the search word and account features of the user account.

[0089] In this step, the first bipartite graph structure can be subjected to feature extraction by using the first graph neural network to obtain the account features of the user account. The graph neural network can model the bipartite graph and capture the internal dependency relationship between the data in the bipartite graph structure. Since the user account is the construction data of the second bipartite graph structure, the account features of the user account can be extracted from the second bipartite graph structure.

[0090] In S14, the first bipartite graph structure is subjected to feature extraction to obtain search features of the search word and data features of the multimedia data.

[0091] In this step, the first bipartite graph structure can be subjected to feature extraction by using the second graph neural network to obtain the data features of the multimedia data and the search features of the search word. Since the search word and the multimedia data are the construction data of the second bipartite graph structure, the search features of the search word and the data features of the multimedia data can be extracted from the second bipartite graph structure.

[0092] The second graph neural network can be obtained based on the training of the first graph neural network, and the learning rate of the second graph neural network is less than that of the first graph neural network, that is, the first bipartite graph and the second bipartite graph can adopt the same graph neural network structure. In the training process of the first graph neural network and the second graph neural network, the neural network parameters of the two can be mutually referenced and promoted, which is conducive to improving the accuracy and recall rate of the classification of the multimedia data and the user account.

[0093] Specifically, in the training process, a first training bipartite graph between the search word and the multimedia data in the training sample and a second training bipartite graph between the search word and the user account in the training sample are constructed.

[0094] Then, the preset graph neural network is trained based on the second training bipartite graph to obtain a first graph neural network, wherein the output of the first graph neural network includes account features of user accounts in the training samples and initial features of search terms.

[0095] Then, the learning rate of the model training is reduced, the first graph neural network is continued to be trained based on the first training bipartite graph, and the model parameters of the first graph neural network are iteratively adjusted to obtain the second graph neural network, wherein the output of the second graph neural network includes the data features of the multimedia data in the training samples and the search features of the search terms.

[0096] It can be understood that since the second graph neural network is trained based on the first graph neural network, the search features of the search terms output by the second graph neural network are more accurate than the initial features of the search terms output by the first graph neural network. Therefore, in the present disclosure, it is not necessary to obtain the initial features of the search terms, and it is only necessary to perform subsequent classification based on the search features of the search terms extracted from the first bipartite graph.

[0097] In S15 , multimedia data, user accounts and search terms are clustered according to the account features, search features and data features to obtain multiple clusters, each of which includes at least one multimedia data, search term or user account.

[0098] In this step, clustering can be performed based on the similarity between account features, search features and data features. For example, the cosine similarity algorithm can be used to calculate the similarity between account features, update features and data features. After obtaining the similarity, any one or more clustering methods such as knn (k-NearestNeighbor) algorithm, k-means (K means) algorithm, etc. can be adopted. This disclosure does not limit this.

[0099] In the present disclosure, after multimedia data, user accounts and search terms are clustered according to account features, search features and data features to obtain multiple clusters, each cluster can be distinguished according to the multimedia data, user accounts and search terms included in each cluster.

[0100] For example, a cluster whose included multimedia data, search terms or user accounts meet preset conditions can be determined as a target cluster, the multimedia data in the target cluster is the target multimedia data, and the user accounts in the target cluster are the target user accounts.

[0101] The preconditions include any one or more of the following:

[0102] The search terms in the cluster are all preset target search terms or contain preset fields. For example, the preset fields can be keywords such as "method", "what to do", and "how"; the preset target user accounts in the cluster account for more than half of the total number of user accounts in the cluster; the amount of preset target multimedia data in the cluster accounts for more than half of the total number of multimedia data in the cluster, or the amount of preset target multimedia data in the cluster is greater than the preset number threshold.

[0103] In this way, if any cluster meets any of the above conditions, the multimedia data, search terms and user accounts in the cluster can all be determined as target multimedia data and target user accounts, thereby quickly and efficiently identifying target multimedia data and target user accounts from a large amount of multimedia data and a large number of user accounts.

[0104] As can be seen from the above, in the technical solution provided by the embodiment of the present disclosure, by constructing a first bipartite graph between search terms and multimedia data and a second bipartite graph between search terms and user accounts, and extracting features of multimedia data and user accounts based on the first bipartite graph and the second bipartite graph, the multimedia data and user accounts are clustered. On the one hand, the number of prior labels required for classification is reduced and the accuracy of classification is improved. On the other hand, in the process of extracting search term features, the information of multimedia data and user accounts is simultaneously utilized, so that the information of multimedia data and user accounts is effectively coordinated and promoted.

[0105] Secondly, an information classification device provided by an embodiment of the present disclosure will be introduced in detail.

[0106] like Figure 2 As shown, an information classification device applied to a server provided by an embodiment of the present disclosure includes:

[0107] An acquisition unit 201 is configured to acquire behavior data of a search term and search results for the search term, wherein the search results include multimedia data and a user account;

[0108] A construction unit 202 is configured to construct a first bipartite graph between the search term and the multimedia data and a second bipartite graph between the search term and the user account according to the search term and the behavior data;

[0109] A first extraction unit 203 is configured to perform feature extraction on the second bipartite graph structure to obtain account features of the user account;

[0110] A second extraction unit 204 is configured to perform feature extraction on the first bipartite graph to obtain search features of the search term and data features of the multimedia data;

[0111] The classification unit 205 is configured to perform clustering of the multimedia data, the user account and the search term based on the account characteristics, the search characteristics and the data characteristics to obtain multiple clusters, each cluster including at least one of the multimedia data, the search term or the user account.

[0112] In one implementation, the classification unit 205 is configured to perform:

[0113] A cluster including the multimedia data, the search terms or the user accounts meeting preset conditions is determined as a target cluster, the multimedia data in the target cluster is the target multimedia data, and the user accounts in the target cluster are the target user accounts.

[0114] In one implementation, the preset conditions include any one or more of the following:

[0115] The search terms in the cluster are all preset target search terms or contain preset fields;

[0116] The preset target user accounts in the cluster account for more than half of the total number of user accounts in the cluster;

[0117] The amount of the preset target multimedia data in the cluster accounts for more than half of the total amount of multimedia data in the cluster, or the amount of the preset target multimedia data in the cluster is greater than a preset amount threshold.

[0118] In one implementation, the constructing unit 202 is configured to execute:

[0119] Calculating, based on the behavior data, a click-through rate and a satisfaction rate between the search term corresponding to the behavior data and the multimedia data;

[0120] Calculating the relevance between the search term and the multimedia data according to the click-through rate and the satisfaction level;

[0121] When the correlation degree is higher than a preset correlation threshold, a connection is established between the search term and the multimedia data to obtain a first bipartite graph, and a connection is established between the search term and the user account corresponding to the multimedia data to obtain a second bipartite graph.

[0122] In one implementation, the behavior data includes the number of times the user displays, clicks, plays, likes, follows, the number of times the play time exceeds a preset time, and play order parameters of the multimedia data;

[0123] The construction unit 202 is configured to execute:

[0124] Calculating a ratio between the number of clicks and the number of impressions as a click-through rate between the search term corresponding to the behavior data and the multimedia data;

[0125] Calculating the sum of the number of times the playback duration exceeds a preset duration, the number of likes, the number of follows, and the playback order parameter, and using the ratio of the sum to the number of playbacks as the satisfaction between the search term corresponding to the behavior data and the multimedia data;

[0126] The product of the click-through rate and the satisfaction level is calculated as the correlation between the search term and the multimedia data.

[0127] In one implementation, the first extraction unit 203 is configured to execute:

[0128] Using the first graph neural network, extracting features from the second bipartite graph structure to obtain initial features of the search term and account features of the user account;

[0129] The second extraction unit 204 is configured to perform:

[0130] A second graph neural network is used to perform feature extraction on the first bipartite graph structure to obtain data features of the multimedia data and search features of the search terms, wherein the second graph neural network is obtained based on the training of the first graph neural network, and the learning rate of the second graph neural network is less than the learning rate of the first graph neural network.

[0131] As can be seen from the above, in the technical solution provided by the embodiment of the present disclosure, by constructing a first bipartite graph between search terms and multimedia data and a second bipartite graph between search terms and user accounts, and extracting features of multimedia data and user accounts based on the first bipartite graph and the second bipartite graph, the multimedia data and user accounts are clustered. On the one hand, the number of prior labels required for classification is reduced and the accuracy of classification is improved. On the other hand, in the process of extracting search term features, the information of multimedia data and user accounts is simultaneously utilized, so that the information of multimedia data and user accounts is effectively coordinated and promoted.

[0132] In the third aspect, an electronic device provided by an embodiment of the present disclosure will be introduced in detail.

[0133] like Figure 3 As shown, this is a schematic structural diagram of an electronic device provided by an embodiment of the present disclosure.

[0134] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory including instructions, wherein the instructions can be executed by a processor of an electronic device to perform the above method. Alternatively, the computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0135] In an exemplary embodiment, a computer program product is also provided. When the computer program product is executed on a computer, the computer is enabled to implement the above-mentioned information classification method.

[0136] Figure 4 FIG1 is a block diagram of another apparatus 800 according to an exemplary embodiment. For example, the apparatus 800 may be a mobile phone, a computer, a digital broadcast electronic device, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0137] Reference Figure 4 , the apparatus 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input / output (I / O) interface 812 , a sensor component 814 , and a communication component 816 .

[0138] The processing component 802 generally controls the overall operation of the device 800, such as operations associated with display, phone calls, data communications, camera operation, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to perform all or part of the steps of the above-described method. In addition, the processing component 802 may include one or more modules to facilitate interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate interaction between the multimedia component 808 and the processing component 802.

[0139] The memory 804 is configured to store various types of data to support operations on the device 800. Examples of such data include instructions for any application or method operating on the device 800, contact data, phone book data, messages, pictures, multimedia data, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0140] The power supply component 807 provides power to the various components of the device 800. The power supply component 807 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device 800.

[0141] The multimedia component 808 includes a screen that provides an output interface between the device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen can be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor can not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the device 800 is in an operating mode, such as a shooting mode or a multimedia data mode, the front camera and / or the rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have a focal length and optical zoom capability.

[0142] The audio component 810 is configured to output and / or input audio signals. For example, the audio component 810 includes a microphone (MIC), which is configured to receive external audio signals when the device 800 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signal can be further stored in the memory 804 or transmitted via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting audio signals.

[0143] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as a keyboard, click wheel, buttons, etc. These buttons may include but are not limited to: a home button, volume buttons, a start button, and a lock button.

[0144] The sensor component 814 includes one or more sensors for providing status assessments for various aspects of the device 800. For example, the sensor component 814 can detect an open / closed position of the device 800, relative positioning of components, such as a display and keypad of the device 800, a change in position of the device 800 or a component of the device 800, presence or absence of user contact with the device 800, orientation or acceleration / deceleration of the device 800, and temperature changes of the device 800. The sensor component 814 can include proximity sensor(s) configured to detect presence of nearby objects without any physical contact. The sensor component 814 can also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor component 814 can also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0145] The communication component 816 is configured to facilitate wired or wireless communication between the device 800 and another device. The device 800 can access a wireless network based on a communication standard, such as WiFi, a cellular network (e.g., 2G, 3G, 4G, or 5G), or a combination thereof. In an example embodiment, the communication component 816 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In an example embodiment, the communication component 816 can further include a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on Radio Frequency Identification (RFID) techniques, infrared data association (IrDA) techniques, ultra-wideband (UWB) techniques, Bluetooth (BT) techniques, and other techniques.

[0146] In an example embodiment, the device 800 can be implemented with one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic elements to perform any of the above-described information classification methods.

[0147] In an example embodiment, a non-transitory computer-readable storage medium including instructions, such as the memory 804 including instructions, is also provided, which can be executed by the processor 820 of the device 800 to complete the above-described methods. Alternatively, the storage medium can be a non-transitory computer-readable storage medium, such as a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disc, and an optical data storage device, and the like.

[0148] According to a fourth aspect of an embodiment of the present disclosure, a storage medium is provided. When instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform any of the above-mentioned information classification methods.

[0149] According to a fifth aspect of an embodiment of the present disclosure, there is provided a computer program product comprising instructions, which, when executed on a computer, enables the computer to implement any of the above-mentioned information classification methods.

[0150] As can be seen from the above, in the technical solution provided by the embodiment of the present disclosure, by constructing a first bipartite graph between search terms and multimedia data and a second bipartite graph between search terms and user accounts, and extracting features of multimedia data and user accounts based on the first bipartite graph and the second bipartite graph, the multimedia data and user accounts are clustered. On the one hand, the number of prior labels required for classification is reduced and the accuracy of classification is improved. On the other hand, in the process of extracting search term features, the information of multimedia data and user accounts is simultaneously utilized, so that the information of multimedia data and user accounts is effectively coordinated and promoted.

[0151] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the following claims.

[0152] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. An information classification method, characterized in that: The method comprises: Obtaining behavioral data of a search term and search results for the search term, wherein the search results include multimedia data and a user account; Based on the search term and the behavior data, a first bipartite graph between the search term and the multimedia data and a second bipartite graph between the search term and the user account are constructed, including: calculating the ratio between the number of clicks and the number of impressions as the click-through rate between the search term corresponding to the behavior data and the multimedia data; calculating the sum of the number of times the playback time exceeds a preset time, the number of likes, the number of follows, and a playback order parameter, and taking the ratio of the sum to the number of playbacks as the satisfaction rate between the search term corresponding to the behavior data and the multimedia data; calculating the product of the click-through rate and the satisfaction rate as the correlation between the search term and the multimedia data; and when the correlation is higher than a preset correlation threshold, establishing a connection between the search term and the multimedia data to obtain a first bipartite graph, and establishing a connection between the search term and the user account corresponding to the multimedia data to obtain a second bipartite graph; performing feature extraction on the second bipartite graph structure to obtain account features of the user account; Performing feature extraction on the first bipartite graph to obtain search features of the search term and data features of the multimedia data; The multimedia data, the user account and the search term are clustered according to the account feature, the search feature and the data feature to obtain a plurality of clusters, each cluster including at least one of the multimedia data, the search term or the user account.

2. The method according to claim 1, characterized in that After clustering the multimedia data, the user account, and the search term according to the account feature, the search feature, and the data feature to obtain a plurality of clusters, the method further includes: A cluster including the multimedia data, the search terms or the user accounts meeting preset conditions is determined as a target cluster, the multimedia data in the target cluster is the target multimedia data, and the user accounts in the target cluster are the target user accounts.

3. The method according to claim 2, characterized in that The preset conditions include any one or more of the following: The search terms in the cluster are all preset target search terms or contain preset fields; The preset target user accounts in the cluster account for more than half of the total number of user accounts in the cluster; The amount of the preset target multimedia data in the cluster accounts for more than half of the total amount of multimedia data in the cluster, or the amount of the preset target multimedia data in the cluster is greater than a preset amount threshold.

4. The method according to claim 1, wherein The extracting features of the second bipartite graph structure to obtain account features of the user account includes: Using the first graph neural network, extracting features from the second bipartite graph structure to obtain account features of the user account; The extracting features of the first bipartite graph to obtain the search features of the search term and the data features of the multimedia data includes: A second graph neural network is used to perform feature extraction on the first bipartite graph structure to obtain data features of the multimedia data and search features of the search terms, wherein the second graph neural network is obtained based on the training of the first graph neural network, and the learning rate of the second graph neural network is less than the learning rate of the first graph neural network.

5. An information classification device, characterized in that: The device comprises: an acquiring unit configured to acquire behavioral data of a search term and a search result for the search term, wherein the search result includes multimedia data and a user account; A construction unit is configured to execute, based on the search term and the behavior data, construction of a first bipartite graph between the search term and the multimedia data and a second bipartite graph between the search term and the user account, including: calculating a ratio between the number of clicks and the number of impressions as a click-through rate between the search term corresponding to the behavior data and the multimedia data; calculating the sum of the number of times the playback time exceeds a preset time, the number of likes, the number of follows, and a playback order parameter, and using the ratio of the sum to the number of playbacks as a satisfaction level between the search term corresponding to the behavior data and the multimedia data; calculating the product of the click-through rate and the satisfaction level as a correlation between the search term and the multimedia data; and establishing a connection between the search term and the multimedia data to obtain a first bipartite graph when the correlation is higher than a preset correlation threshold, and establishing a connection between the search term and the user account corresponding to the multimedia data to obtain a second bipartite graph; a first extraction unit configured to perform feature extraction on the second bipartite graph structure to obtain account features of the user account; a second extraction unit configured to perform feature extraction on the first bipartite graph to obtain search features of the search term and data features of the multimedia data; The classification unit is configured to perform clustering of the multimedia data, the user account and the search term based on the account characteristics, the search characteristics and the data characteristics to obtain multiple clusters, each cluster including at least one of the multimedia data, the search term or the user account.

6. The device according to claim 5, characterized in that The classification unit is configured to perform: A cluster including the multimedia data, the search terms or the user accounts meeting preset conditions is determined as a target cluster, the multimedia data in the target cluster is the target multimedia data, and the user accounts in the target cluster are the target user accounts.

7. The device according to claim 6, characterized in that The preset conditions include any one or more of the following: The search terms in the cluster are all preset target search terms or contain preset fields; The preset target user accounts in the cluster account for more than half of the total number of user accounts in the cluster; The amount of the preset target multimedia data in the cluster accounts for more than half of the total amount of multimedia data in the cluster, or the amount of the preset target multimedia data in the cluster is greater than a preset amount threshold.

8. The device according to claim 5, characterized in that The first extraction unit is configured to execute: Using the first graph neural network, extracting features from the second bipartite graph structure to obtain initial features of the search term and account features of the user account; The second extraction unit is configured to execute: A second graph neural network is used to perform feature extraction on the first bipartite graph structure to obtain data features of the multimedia data and search features of the search terms, wherein the second graph neural network is obtained based on the training of the first graph neural network, and the learning rate of the second graph neural network is less than the learning rate of the first graph neural network.

9. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the information classification method according to any one of claims 1 to 4.

10. A storage medium, characterized in that: When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the information classification method according to any one of claims 1 to 4.

11. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the information classification method according to any one of claims 1 to 4 is implemented.

Images