Method, device and computer program product for determining a resource to be recommended

Abstract

The present disclosure provides a method and device for determining a resource to be recommended, electronic equipment, storage medium and program product, which relates to the technical field of computers, specifically to the technical field of information flow recommendation, and can be applied to the information flow recommendation scene. The specific implementation scheme is as follows: obtaining a plurality of resources corresponding to a target user that are recalled; determining a target resource that is of interest to the target user from the plurality of resources, and exposing the target resource to the target user; determining a trigger from the interactive resources of the target user, wherein the interactive resources include historical recommended resources that the target user has interacted with up to the target resource; and recalling a resource to be recommended for the target user according to the trigger. The present disclosure makes the determined trigger have high confidence, and further improves the recall accuracy and the accuracy of the determined resource to be recommended based on the trigger with high confidence.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of computer, in particular to the technical field of information flow recommendation, and more particularly to a method and apparatus for determining a resource to be recommended, an electronic device, a storage medium, and a computer program product, which can be applied to an information flow recommendation scenario. BACKGROUND

[0002] In the current information flow recommendation system, there are mainly two stages of recall and sorting. The ICF (Item-based Collaborative Filtering) recall method is a very efficient and mature recall way. However, the precision of the ICF recall method needs to be improved. SUMMARY

[0003] The present disclosure provides a method and apparatus for determining a resource to be recommended, an electronic device, a storage medium, and a computer program product.

[0004] According to a first aspect, a method for determining a resource to be recommended is provided, comprising: obtaining a plurality of resources corresponding to a target user and recalled; determining a target resource of interest of the target user from the plurality of resources, and exposing the target resource to the target user; determining a trigger item from an interactive resource of the target user, wherein the interactive resource includes a historical recommended resource that the target user has interacted with up to the target resource; and recalling a resource to be recommended of the target user according to the trigger item.

[0005] According to a second aspect, a device for determining a resource to be recommended is provided, comprising: an obtaining unit configured to obtain a plurality of resources corresponding to a target user and recalled; a first determining unit configured to determine a target resource of interest of the target user from the plurality of resources, and expose the target resource to the target user; a second determining unit configured to determine a trigger item from an interactive resource of the target user, wherein the interactive resource includes a historical recommended resource that the target user has interacted with up to the target resource; and a recalling unit configured to recall a resource to be recommended of the target user according to the trigger item.

[0006] According to a third aspect, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described in any implementation manner of the first aspect.

[0007] According to a fourth aspect, a non-transitory computer-readable storage medium storing computer instructions is provided, and the computer instructions are used to enable a computer to perform the method described in any implementation manner of the first aspect.

[0008] According to a fifth aspect, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the method according to any implementation of the first aspect.

[0009] According to the technology of the present disclosure, a method for determining a resource to be recommended is provided, by obtaining a plurality of resources corresponding to a target user which are recalled; determining a target resource interested by the target user from the plurality of resources, and exposing the target resource to the target user; determining a trigger from the interactive resources of the target user, wherein the interactive resources include historical recommended resources interacted by the target user before the target resource, so that the determined trigger has a high confidence, and further based on the trigger with high confidence, the recall precision and the accuracy of the determined resource to be recommended are improved.

[0010] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present disclosure, nor to limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0011] The accompanying drawings are used to better understand the present scheme, and do not constitute a limitation on the present disclosure. Among them:

[0012] Figure 1 is an exemplary system architecture diagram to which an embodiment according to the present disclosure can be applied;

[0013] Figure 2 is a flowchart of an embodiment of the method for determining a resource to be recommended according to the present disclosure;

[0014] Figure 3 is a schematic diagram of an application scenario of the method for determining a resource to be recommended according to the present embodiment;

[0015] Figure 4 is a flowchart of another embodiment of the method for determining a resource to be recommended according to the present disclosure;

[0016] Figure 5 is a structural diagram of an embodiment of the device for determining a resource to be recommended according to the present disclosure;

[0017] Figure 6 is a structural schematic diagram of a computer system suitable for implementing an embodiment of the present disclosure. DETAILED DESCRIPTION

[0018] Exemplary embodiments of the present disclosure are described herein with reference to the accompanying drawings, which are meant to be exemplary in nature, and include various details intended to facilitate understanding of the present disclosure. However, those skilled in the art will recognize that the embodiments described herein can be practiced with variation of the details as not to depart from the scope and spirit of the present disclosure. Accordingly, the present disclosure is not limited to the embodiments described herein, but rather the scope of the present disclosure is to be accorded the broadest scope of the claims, and any and all equivalents thereof.

[0019] In the technical solutions of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information comply with relevant laws and regulations and do not violate public order and good customs.

[0020] Figure 1 An exemplary architecture 100 to which the method and apparatus for determining a resource to be recommended according to the present disclosure can be applied is shown.

[0021] As shown in Figure 1 The system architecture 100 can include terminal devices 101, 102, 103, a network 104 and a server 105. The terminal devices 101, 102, 103 are communicatively connected to form a topology network, and the network 104 is used as a medium to provide a communication link between the terminal devices 101, 102, 103 and the server 105. The network 104 can include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0022] The terminal devices 101, 102, 103 can be hardware devices or software that support network connection to interact and process data. When the terminal devices 101, 102, 103 are hardware, they can be various electronic devices that support network connection, information acquisition, interaction, display, processing, etc., including but not limited to smartphones, tablet computers, e-book readers, laptop computers and desktop computers, etc. When the terminal devices 101, 102, 103 are software, they can be installed in the above-mentioned electronic devices. They can be implemented as multiple software or software modules for providing distributed services, or as a single software or software module. No specific limitation is made herein.

[0023] The server 105 can be a server that provides various services, such as receiving network requests from the terminal devices 101, 102, 103, and determining the resource to be recommended for the target user according to the determined high-confidence trigger item through a preset recall method. As an example, the server 105 can be a cloud server.

[0024] It should be noted that the server can be hardware or software. When the server is hardware, it can be implemented as a distributed server cluster composed of multiple servers, or as a single server. When the server is software, it can be implemented as multiple software or software modules (for example, software or software modules used to provide distributed services), or as a single software or software module. This is not specifically limited here.

[0025] It should also be noted that the determination method of the resource to be recommended provided by the embodiments of the present disclosure can be executed by a server, or by a terminal device, or by the server and the terminal device in cooperation with each other. Accordingly, the determination apparatus of the resource to be recommended can include various parts (for example, various units), which can all be arranged in the server, or all be arranged in the terminal device, or be arranged in the server and the terminal device respectively.

[0026] It should be understood that Figure 1 The number of terminal devices, networks and servers in the system architecture is only illustrative. According to the needs of implementation, there can be any number of terminal devices, networks and servers. When the electronic device on which the determination method of the resource to be recommended runs does not need to perform data transmission with other electronic devices, the system architecture can only include the electronic device (for example, a server or a terminal device) on which the determination method of the resource to be recommended runs.

[0027] Please refer to Figure 2 , Figure 2 A flowchart of a determination method of a resource to be recommended provided by an embodiment of the present disclosure is shown in FIG. 2. In flow 200, the following steps are included:

[0028] In step 201, a plurality of resources corresponding to a target user that are recalled are obtained.

[0029] In this embodiment, the execution subject (for example, a terminal device or a server) of the determination method of the resource to be recommended can obtain the plurality of resources corresponding to the target user that are recalled from a remote place through a wired network connection or a wireless network connection, or from a local place. Figure 1

[0030] Among them, the resource represents digital media data in the information age. In different application scenarios, the resource can be specifically represented as different data. As an example, in the video recommendation scenario, the resource can be short video, long video and the like; in the news recommendation scenario, the resource can be news, articles, news videos and the like; in the advertisement recommendation scenario, the resource can be entertainment news, political news and the like.

[0031] In this embodiment, the above execution subject can use a preset recall algorithm to recall the plurality of resources corresponding to the target user from a resource library.

[0032] ​As an example, the preset recall algorithm is an ICF recall algorithm. Specifically, first, a user-resource graph model is constructed, and an embeding vector of the resource is obtained through a graph embeding method. Then, the cosine similarity between each two resources is calculated based on the embeding vector of the resource. The greater the similarity value, the more similar the two resources. Finally, an index is constructed for online recall. For the index, the following examples are given:

[0033] nid1 [nid2 & 0.91, nid3 & 0.82, nid4 & 0.71,...]

[0034] nid2 [nid1 & 0.91, nid4 & 0.85, nid5 & 0.67,...]

[0035] Wherein, each row is an index; the first column is the trigger (for example, nid1 in the first row, and nid2 in the second row); the second column is the information of multiple resources. For example, in the second column of the first row, "nid2 & 0.91" indicates that the cosine similarity between nid2 and nid1 is 0.91, "nid3 & 0.82" indicates that the cosine similarity between nid3 and nid1 is 0.82. And in the second column, the multiple resources are sorted according to the order of the cosine similarity from large to small. Generally, only the top 50 resources with the highest cosine similarity are retained, and the cosine similarity of the retained resources needs to be greater than a preset similarity threshold. The preset similarity threshold can be set according to actual conditions, which is not limited here.

[0036] In the process of recalling multiple resources corresponding to the target user, the latest multiple resources interacted by the user (for example, clicked, browsed) are obtained first, and then the latest multiple resources are used as triggers to access the index to recall the resources similar to the triggers.

[0037] As another example, in the process of continuously recommending the target user by the method represented by steps 201-204, the multiple resources corresponding to the target user obtained by step 201 in this recommendation process are the to-be-recommended resources recalled by the trigger determined by step 204 in the last recommendation operation according to the interactive resources of the target user.

[0038] In some optional implementations of the embodiment, the above execution subject can execute the above step 201 by the following method: obtaining multiple resources corresponding to the target user recalled by multiple different recall methods.

[0039] In addition to the above ICF recall method, the multiple different recall methods can also include the following methods:

[0040] Content-based Recall method: This method is based on the content features of items to perform recall. By calculating the content similarity of the items involved in the user's historical behavior, similar items to the items the user liked in the past are selected from the candidate items to perform recall.

[0041] Matrix Factorization Recall method: Matrix factorization is a classic collaborative filtering method that maps users and items to a latent low-dimensional space by decomposing the user-item rating matrix, thereby discovering the underlying factors hidden behind user behavior. Using the results of matrix factorization, recall recommendations can be made.

[0042] Tag-based Recall method: This method uses the label information that users give to items to perform recall. By analyzing the labels that users give to items, items with the same or similar labels are found to perform recall. The labels can be added by the user themselves or from other sources.

[0043] Knowledge Graph-based Recall method: Knowledge graph contains a large number of entities and relationships between entities, and the structure information of knowledge graph can be used in the recommendation system to perform recall. By matching the user's historical behavior data with the entities and relationships in the knowledge graph, items that may be of interest are found to perform recall.

[0044] In this implementation, first, for each of the multiple different recall methods, the execution subject can recall multiple resources corresponding to the target user from the resource library through the recall method; then, the resources recalled by each recall method are fused, and operations such as deduplication are performed, and finally, multiple resources fused by multiple different recall methods are obtained.

[0045] In this implementation, multiple different recall methods are used to perform recall operations to obtain multiple resources corresponding to the target user, which improves the richness of the resources and further helps to improve the confidence of the trigger item determined from the multiple resources in the subsequent steps.

[0046] Step 202: Determine the target resource of interest to the target user from the multiple resources, and expose the target resource to the target user.

[0047] In this embodiment, the execution subject can determine the target resource of interest to the target user from the multiple resources, and expose the target resource to the target user.

[0048] As an example, the execution subject can randomly determine a plurality of target resources of interest to the target user from a plurality of resources, and push the plurality of target resources to a target application corresponding to the target user (for example, a news application corresponding to a news resource, a short video application corresponding to a short video resource), so as to expose the plurality of target resources.

[0049] As another example, the execution subject can determine the degree of interest of the target user to each resource through an interest determination model, and sort the resources according to the degree of interest from large to small, determine the first specified number of resources in the front as target resources, and expose the target resources to the target user. Wherein, the interest determination model is used to represent the degree of interest between the target user and the resource, and can be obtained by training a neural network model. The first specified number can be specifically set according to actual conditions, which is not limited here.

[0050] In some optional implementations of the embodiment, the execution subject can perform the above step 202 in the following manner:

[0051] First, a plurality of resources are multi-target sorted to determine the sorting information of the plurality of resources.

[0052] Wherein, the multi-target sorting represents sorting the plurality of resources according to a plurality of sorting indicators.

[0053] As an example, the execution subject can input a plurality of resources into a multi-target sorting model, so as to obtain sorting basis information (such as score) for the sorting process. Further, the plurality of resources are sorted according to the sorting basis information of the plurality of resources to obtain the sorting information of the plurality of resources.

[0054] Wherein, the plurality of sorting indicators may, for example, be estimated click rate, estimated duration, estimated like, etc. It should be noted that the plurality of sorting indicators in different recommendation scenarios may be different. In the present implementation, the execution subject can determine the plurality of sorting indicators corresponding to the recommendation scenario according to the specific recommendation scenario. As an example, the execution subject or the electronic device in communication connection with the execution subject is provided with a two-dimensional table representing the correspondence between the recommendation scenario and the sorting indicator, so as to determine the plurality of sorting indicators corresponding to the recommendation scenario through the two-dimensional table.

[0055] Second, the target resource is determined from the plurality of resources according to the sorting information.

[0056] In the present implementation, the execution subject can select the first specified number of resources in the front as target resources according to the sorting information of the plurality of resources.

[0057] In this implementation, the multiple resources are sorted in a multi-target sorting manner. By comprehensively considering multiple sorting indexes, the accuracy of the sorting information is improved, and the accuracy of the determined target resource is further improved, which is further conducive to improving the confidence of the trigger item determined from the target resource.

[0058] In some optional implementations of the embodiment, the execution subject can perform the first step in the following manner.

[0059] First, the predicted scores of the multiple resources under each sorting index are determined. The predicted score is used to represent the interest degree of the target user in the resource.

[0060] For example, taking the estimated click rate, the estimated duration, and the estimated like as the multiple sorting indexes, the predicted score q ctr under the estimated click rate index, the predicted score q dur under the estimated duration index, and the predicted score q like under the estimated like index can be determined.

[0061] Then, for each resource in the multiple resources, the fusion score corresponding to the resource is determined according to the predicted score of the resource under each sorting index.

[0062] In this implementation, the execution subject can preset the fusion manner, and then determine the fusion score corresponding to the resource based on the preset fusion manner.

[0063] For example, the execution subject can perform score fusion in the following manner:

[0064] score=(q ctr ) a ×(q dur ) b ×(q like ) c

[0065] wherein score represents the fusion score, and a, b, and c represent preset hyperparameters.

[0066] Finally, the multiple resources are sorted according to the fusion scores to determine the sorting information of the multiple resources.

[0067] In this implementation, the multiple resources can be sorted according to the order of the fusion scores from large to small to determine the first specified number of target resources in the front.

[0068] In the present implementation, the fusion scores of the first specified number of target resources in the front of the ranking can be logged in the log, and the format of the logging can be, for example, user1_resource1_fusion score1, user1_resource2_fusion score2, …, user1_resourceM_fusion scoreM. In the subsequent processing steps, the fusion scores of the resources can be determined from the logging in the log to determine the degree of interest of the target user pair.

[0069] In the present implementation, the predicted scores under each ranking index are fused to obtain the fusion score, and then the ranking is performed according to the fusion score, which further improves the accuracy of the ranking information.

[0070] In step 203, the trigger item is determined from the interactive resource of the target user.

[0071] In the present embodiment, the above execution subject can determine the trigger item from the interactive resource of the target user. The interactive resource includes the historical recommended resource that the target user has interacted with up to the target resource.

[0072] As an example, after the target resource is exposed to the target user, the interactive behavior of the target user with the target resource can be monitored to determine the resources that the target user has interacted with, and the resources that the target user has interacted with in the target resource are fused with the resources that the target user has interacted with before to obtain the interactive resource. For example, it is determined that the target user has interacted with 3 resources in the target resource, and the target user has interacted with 100 resources within one month before the target resource is exposed to the target user, and finally an interactive resource including 103 resources is obtained.

[0073] After determining the interactive resource of the target user, the above execution subject can determine the trigger item required by the preset recall method therefrom. As an example, for each resource in the interactive resource, the above execution subject can determine the actual degree of interest of the target user in the resource according to the specific parameters (such as the viewing time, the page turning operation) in the interactive behavior of the target user with the resource; and then the interactive resource is ranked according to the actual degree of interest, and the second specified number of interactive resources in the front of the ranking are determined as the trigger item. The second specified number can be specifically set according to the actual situation, which is not limited herein.

[0074] As another example, in order to reduce the information processing amount in the information recommendation process, the above execution subject can rank the interactive resource according to the predicted degree of interest (specifically, for example, the fusion score) in the process of determining the target resource, and determine the second specified number of interactive resources in the front of the ranking as the trigger item.

[0075] In some optional implementations of the present embodiment, the above execution subject can perform the above step 203 in the following manner:

[0076] First, a user model corresponding to the target user is obtained.

[0077] The user model includes historical recommended resources that the target user has interacted with before the target resource.

[0078] Generally, the execution subject may have performed multiple resource recommendation operations on the target user before exposing the target resource. The execution subject can generate a user model corresponding to the target user according to the historical recommended resources that the target user has interacted with before the target resource.

[0079] In order to ensure the timeliness of the interaction resources in the user model, the interaction resources in the user model can be resources with recent interaction times. For example, the interaction resources with a difference between the interaction time and the current time within a preset difference range can be used to generate the user model of the target user. The preset difference range can be set according to actual conditions, for example, the preset difference range is one month.

[0080] Second, the user model is updated according to the resources that the target user has interacted with in the target resource, to obtain an updated user model.

[0081] In the implementation mode, the execution subject can supplement the resources that the target user has interacted with in the target resource into the user model to obtain an updated user model corresponding to the target user.

[0082] Third, a trigger item is determined from the historical recommended resources included in the updated user model.

[0083] As an example, the execution subject can select a preset number of resources with the greatest interest degree of the target user from the historical recommended resources included in the updated user model as the trigger item. The interest degree can be represented by the fusion score.

[0084] In the implementation mode, a user model corresponding to each user is set to represent the interaction resources of the user, and the trigger item can be determined according to the updated user model, so that the information processing can be performed for each user, and the effectiveness and efficiency of the information processing process are improved.

[0085] In some optional implementation modes of the embodiment, the execution subject can perform the third step in the following manner:

[0086] First, a preset number of initial trigger items are determined from the historical recommended resources included in the updated user model according to the negative correlation between the time difference and the priority.

[0087] The time difference represents the difference between the interaction time of the target user and the historical recommended resource and the current time.

[0088] In this implementation, the preset number of initial trigger items closest to the current time in the interaction time are determined from the historical recommended resources included in the updated user model.

[0089] Then, the trigger item is determined from the preset number of initial trigger items according to the degree of interest of the target user in the historical recommended resources.

[0090] In this implementation, the execution subject determines the historical recommended resources with the highest degree of interest of the user in the initial trigger items as the trigger items. The degree of interest may be determined by fusing the scores, for example.

[0091] In this implementation, the initial trigger items with high timeliness are first determined according to the time difference, and then the trigger items are selected therefrom, thereby improving the timeliness of the trigger items while ensuring the confidence of the trigger items.

[0092] In some optional implementations of this embodiment, the execution subject may perform the operation of determining the trigger items from the preset number of initial trigger items in the following manner:

[0093] In response to determining that the number of target historical recommended resources in the updated user model exceeds the preset number, the preset number of initial trigger items are determined from the target historical recommended resources in the updated user model according to the negative correlation between the time difference and the priority.

[0094] The target historical recommended resources are the historical recommended resources corresponding to the time difference less than the preset time difference threshold.

[0095] In this implementation, the preset time difference threshold may be set according to actual conditions, for example, the preset time difference threshold is one week. By limiting the preset time difference threshold, the timeliness of the initial trigger items is further improved.

[0096] In this implementation, in response to determining that the number of target historical recommended resources in the updated user model does not exceed the preset number, all target historical recommended resources are determined as initial trigger items.

[0097] In step 204, the target user's to-be-recommended resources are recalled according to the trigger items.

[0098] In this embodiment, the execution subject may recall the target user's to-be-recommended resources according to the trigger items.

[0099] Specifically, the execution subject may use a preset recall method to recall the target user's to-be-recommended resources from the resource library according to the trigger items.

[0100] Taking the preset recall method as the ICF recall method as an example, resources similar to the trigger item are recalled from the index representing the similarity between the trigger item and the resources as the target user's corresponding to-be-recommended resources.

[0101] In actual recommendation scenarios, multiple recommendation operations are often performed on the target user. For example, in a short video recommendation scenario, based on the target user's recommendation request, after multiple short videos are recommended to the target user, the target user will watch and like the recommended short videos or directly swipe away the recommended short videos; after watching the recommended short videos this time, the target user will initiate a recommendation request again to obtain short videos recommended by the short video platform again.

[0102] In the multiple recommendation scenarios, the steps 201-204 are executed in a loop to continuously recommend resources to the target user. Specifically, in one recommendation operation, the execution subject executes the steps 201-204 to determine the target user's to-be-recommended resources. After determining the to-be-recommended resources, the execution subject can need to recommend resources to the target user, which is essentially the process of determining target resources and exposing target resources as described in steps 201-202. That is, steps 201-202 in the next recommendation operation. Essentially, the to-be-recommended resources obtained in step 204 in the last recommendation operation can be regarded as the target user's corresponding multiple resources obtained in step 201 in the current recommendation operation.

[0103] Continuing to refer to Figure 3 , Figure 3 is one of the application scenarios of the to-be-recommended resource determination method according to the present embodiment. In the application scenario of Figure 3 , the user 301 browses short video resources through a short video application on the terminal device 302. The server 303 providing services to the short video application first obtains the recalled multiple resources corresponding to the target user; then, determines the target resources of interest to the target user from the multiple resources, and exposes the target resources to the target user, monitors the target user's interaction behavior, and determines the resources in the target resources that the target user has interacted with; then, fuses the interaction resources before the target resources and the resources in the target resources that the target user has interacted with to obtain the interaction resources, and further determines the trigger item from the target user's interaction resources; finally, according to the trigger item, the to-be-recommended resources of the target user are recalled.

[0104] In this embodiment, a method for determining a resource to be recommended is provided. The method includes: obtaining a plurality of resources corresponding to a target user and recalled by a plurality of different recall methods; determining a target resource of interest to the target user from the plurality of resources and exposing the target resource to the target user; determining a trigger item from an interactive resource of the target user, wherein the interactive resource includes a historical recommended resource that the target user has interacted with up to the target resource, so that the determined trigger item has a high confidence, and further based on the trigger item with the high confidence, the recall accuracy and the accuracy of the determined resource to be recommended are improved.

[0105] With reference to Figure 4 , another embodiment of a method for determining a resource to be recommended according to the present disclosure is shown. In the flow 400, the following steps are included:

[0106] Step 401, obtaining a plurality of resources corresponding to a target user and recalled by a plurality of different recall methods.

[0107] Step 402, determining a predicted score of each of a plurality of ranking indicators for a plurality of resources.

[0108] The predicted score is used to represent the degree of interest of the target user to the resource.

[0109] Step 403, for each of the plurality of resources, determining a fusion score corresponding to the resource according to the predicted score of the resource under each ranking indicator.

[0110] Step 404, ranking the plurality of resources according to the fusion score to determine ranking information of the plurality of resources.

[0111] Step 405, determining a target resource from the plurality of resources according to the ranking information and exposing the target resource to the target user.

[0112] Step 406, obtaining a user model corresponding to the target user.

[0113] The user model includes a historical recommended resource that the target user has interacted with before the target resource.

[0114] Step 407, updating the user model according to the resource that the target user has interacted with in the target resource to obtain an updated user model.

[0115] Step 408, in response to determining that the number of target historical recommended resources in the updated user model exceeds a preset number, determining a preset number of initial trigger items from the target historical recommended resources in the updated user model according to the negative correlation between time difference and priority.

[0116] The target historical recommended resource is a historical recommended resource corresponding to a time difference less than a preset time difference threshold.

[0117] At step 409, the trigger item is determined from the preset number of initial trigger items according to the fusion score of the historical recommended resource in the initial trigger item.

[0118] At step 410, the to-be-recommended resource of the target user is recalled according to the trigger item.

[0119] It should be noted that the above steps 401-410 are cyclically executed in the process of continuously recommending the target user for multiple times. The to-be-recommended resource determined at step 410 in the last recommendation process is the multiple resources required to be obtained at step 401 in the current recommendation process. In this way, the to-be-recommended resource of the target user is determined continuously.

[0120] As can be seen from the embodiment, compared with the corresponding embodiment, the flow 400 of the determination method of the to-be-recommended resource in the embodiment specifically describes the process of determining the target resource and the process of determining the trigger item, improves the confidence of the trigger item on the basis of ensuring the timeliness of the trigger item, and further helps to improve the accuracy of the to-be-recommended resource. Figure 2

[0121] With reference to the above-mentioned method, the present disclosure provides an embodiment of a to-be-recommended resource determination device. Figure 5 The device embodiment corresponds to the method embodiment shown in Figure 2 , and the device can be specifically applied to various electronic devices.

[0122] As shown in Figure 5 , the to-be-recommended resource determination device 500 includes: an acquisition unit 501 configured to acquire a plurality of resources corresponding to a target user recalled; a first determination unit 502 configured to determine a target resource interested by the target user from the plurality of resources and expose the target resource to the target user; a second determination unit 503 configured to determine a trigger item from an interactive resource of the target user, wherein the interactive resource includes a historical recommended resource interacted by the target user before the target resource; and a recall unit 504 configured to recall a to-be-recommended resource of the target user according to the trigger item.

[0123] In some optional implementations of the embodiment, the first determination unit 502 is further configured to: perform multi-target sorting on the plurality of resources to determine sorting information of the plurality of resources, wherein the multi-target sorting represents sorting the plurality of resources according to a plurality of sorting indicators; and determine the target resource from the plurality of resources according to the sorting information.

[0124] ​In some optional implementations of the present embodiment, the first determining unit 502 is further configured to: determine a predicted score of each of the plurality of resources under each of the plurality of ranking indexes, where the predicted score is used to represent the degree of interest of the target user in the resource; for each of the plurality of resources, determine a fusion score corresponding to the resource according to the predicted score of the resource under each of the plurality of ranking indexes; and rank the plurality of resources according to the fusion scores to determine the ranking information of the plurality of resources.

[0125] In some optional implementations of the present embodiment, the second determining unit 503 is further configured to: obtain a user model corresponding to the target user, where the user model includes historical recommended resources that the target user has interacted with before the target resource; update the user model according to the resource that the target user has interacted with in the target resource to obtain an updated user model; and determine the trigger from the historical recommended resources included in the updated user model.

[0126] In some optional implementations of the present embodiment, the second determining unit 503 is further configured to: determine a preset number of initial triggers from the historical recommended resources included in the updated user model according to a negative correlation between a time difference and a priority, where the time difference represents a difference between an interaction time between the target user and the historical recommended resource and a current time; and determine the trigger from the preset number of initial triggers according to the degree of interest of the target user in the historical recommended resource.

[0127] In some optional implementations of the present embodiment, the second determining unit 503 is further configured to: in response to determining that the number of target historical recommended resources in the updated user model exceeds a preset number, determine a preset number of initial triggers from the target historical recommended resources in the updated user model according to a negative correlation between a time difference and a priority, where the target historical recommended resource is a historical recommended resource corresponding to a time difference less than a preset time difference threshold.

[0128] In some optional implementations of the present embodiment, the obtaining unit 501 is further configured to: obtain a plurality of resources corresponding to the target user that are recalled by a plurality of different recall methods.

[0129] In the present embodiment, a determination apparatus for a to-be-recommended resource is provided. The apparatus obtains a plurality of resources corresponding to a target user that are recalled. The apparatus determines a target resource of interest of the target user from the plurality of resources and exposes the target resource to the target user. The apparatus determines a trigger from an interaction resource of the target user, where the interaction resource includes historical recommended resources that the target user has interacted with up to the target resource. The determined trigger has a high confidence, which improves the recall accuracy and the accuracy of the determined to-be-recommended resource.

[0130] According to an embodiment of the present disclosure, the present disclosure further provides an electronic device, comprising: at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to implement the method for determining a resource to be recommended described in any of the above embodiments.

[0131] According to an embodiment of the present disclosure, the present disclosure further provides a readable storage medium, which stores computer instructions for enabling a computer to implement the method for determining a resource to be recommended described in any of the above embodiments when the computer executes the computer instructions.

[0132] The present disclosure provides a computer program product, which, when executed by a processor, can implement the method for determining a resource to be recommended described in any of the above embodiments.

[0133] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices, and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present disclosure described and / or claimed in this document.

[0134] As shown in Figure 6 The device 600 includes a computing unit 601 that can perform various appropriate actions and processes in accordance with a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. Various programs and data required for operation of the device 600 can also be stored in the RAM 603. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other through a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0135] A number of components in the device 600 are connected to the I / O interface 605, including: an input unit 606, such as a keyboard, a mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a magnetic disk, an optical disk, etc.; and a communication unit 609, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information / data with other devices through computer networks, such as the Internet, and / or various telecommunication networks.

[0136] The computing unit 601 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 601 performs various methods and processes described above, such as the determination method of the resource to be recommended. For example, in some embodiments, the determination method of the resource to be recommended can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as the storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded to the RAM 603 and executed by the computing unit 601, one or more steps of the determination method of the resource to be recommended described above can be performed. Alternatively, in other embodiments, the computing unit 601 can be configured to perform the determination method of the resource to be recommended by any other appropriate means, such as by means of firmware.

[0137] Various implementations of the systems and techniques described above herein can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0138] Program code for carrying out methods of the present disclosure can be written in any combination of one or more programming languages. The program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, produces the functions / operations specified in the flowcharts and / or block diagrams. The program code can be executed entirely on a machine, partially on a machine, partially on a machine as a standalone software package, or entirely on a remote machine or server.

[0139] In the context of the present disclosure, a machine-readable medium can be a tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium will include one or more lines of electrical connections, portable computer disks, hard disk drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0140] To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0141] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0142] The computer system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of large management difficulty and weak business scalability in traditional physical host and virtual private server (VPS, Virtual Private Server) services; or can be a server of a distributed system, or a server combined with a blockchain.

[0143] According to the technical scheme of the embodiment of the present disclosure, a determination method of a to-be-recommended resource is provided, a plurality of resources corresponding to a target user are recalled, a target resource interested by the target user is determined from the plurality of resources, and the target resource is exposed to the target user; a trigger item is determined from an interactive resource of the target user, wherein the interactive resource includes a historical recommended resource that the target user has interacted with up to the target resource, so that the determined trigger item has a high confidence, and then based on the trigger item with high confidence, the recall accuracy and the accuracy of the determined to-be-recommended resource are improved.

[0144] It should be understood that the various forms of flow shown above can be reordered, added to, or deleted from. For example, the steps described in the present disclosure can be executed in parallel, in sequence, or in different orders, as long as the desired results of the technical scheme provided by the present disclosure can be achieved, which is not limited herein.

[0145] The above detailed description does not constitute a limitation on the protection scope of the present disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A recall method of resources to be recommended, comprising: obtaining a plurality of resources corresponding to a target user and recalled; determining a target resource of interest to the target user from the plurality of resources and exposing the target resource to the target user; obtaining a user model corresponding to the target user, wherein the user model comprises historical recommended resources that the target user has interacted with before the target resource; updating the user model according to a resource that the target user has interacted with in the target resource to obtain an updated user model; determining a preset number of initial trigger items from historical recommended resources included in the updated user model according to a negative correlation between a time difference and a priority, wherein the time difference represents a difference between an interaction time between the target user and the historical recommended resource and a current time; determining a trigger item from the preset number of initial trigger items according to a degree of interest of the target user to the historical recommended resource; recalling resources to be recommended for the target user according to the trigger item.

2. The method of claim 1, wherein, The determining of the target resource of interest to the target user from the plurality of resources comprises: performing multi-objective sorting on the plurality of resources to determine sorting information of the plurality of resources, wherein the multi-objective sorting represents sorting the plurality of resources according to a plurality of sorting indicators; determining the target resource from the plurality of resources according to the sorting information.

3. The method of claim 2, wherein, The multi-objective sorting of the plurality of resources to determine the sorting information of the plurality of resources comprises: determining a predicted score of the plurality of resources under each sorting indicator in the plurality of sorting indicators, wherein the predicted score is used to represent a degree of interest of the target user to a resource; for each resource in the plurality of resources, determining a fusion score corresponding to the resource according to the predicted score of the resource under each sorting indicator; sorting the plurality of resources according to the fusion score to determine the sorting information of the plurality of resources.

4. The method of claim 1, wherein, The determining of the preset number of initial trigger items from the historical recommended resources included in the updated user model according to the negative correlation between the time difference and the priority comprises: in response to determining that a number of target historical recommended resources in the updated user model exceeds the preset number, determining the preset number of initial trigger items from the target historical recommended resources in the updated user model according to the negative correlation between the time difference and the priority, wherein the target historical recommended resource is a historical recommended resource corresponding to a time difference less than a preset time difference threshold.

5. The method of claim 1, wherein, The obtaining of the plurality of resources corresponding to the target user and recalled comprises: obtaining the plurality of resources corresponding to the target user and recalled by a plurality of different recall methods.

6. A determination apparatus of resources to be recommended, comprising: an obtaining unit configured to obtain a plurality of resources corresponding to a target user and recalled; a first determining unit configured to determine a target resource of interest to the target user from the plurality of resources and expose the target resource to the target user; The second determining unit is configured to: obtain a user model corresponding to the target user, wherein the user model comprises historical recommended resources that the target user has interacted with before the target resource; update the user model according to the resource that the target user has interacted with in the target resource, to obtain an updated user model; determine a preset number of initial trigger items from the historical recommended resources included in the updated user model according to a negative correlation between a time difference and a priority, wherein the time difference represents a difference between an interaction time between the target user and the historical recommended resource and a current time; and determine a trigger item from the preset number of initial trigger items according to an interest degree of the target user in the historical recommended resource. The recalling unit is configured to recall the to-be-recommended resource of the target user according to the trigger item.

7. The apparatus of claim 6, wherein, The first determining unit is further configured to: perform multi-target sorting on the plurality of resources to determine sorting information of the plurality of resources, wherein the multi-target sorting represents sorting the plurality of resources according to a plurality of sorting indexes; and determine the target resource from the plurality of resources according to the sorting information.

8. The apparatus of claim 7, wherein, The first determining unit is further configured to: determine a predicted score of each sorting index in the plurality of sorting indexes for the plurality of resources, wherein the predicted score is used to represent an interest degree of the target user in a resource; for each resource in the plurality of resources, determine a fusion score corresponding to the resource according to the predicted score of each sorting index of the resource; and sort the plurality of resources according to the fusion score to determine the sorting information of the plurality of resources.

9. The apparatus of claim 6, wherein, The second determining unit is further configured to: in response to determining that the number of target historical recommended resources in the updated user model exceeds the preset number, determine the preset number of initial trigger items from the target historical recommended resources in the updated user model according to a negative correlation between a time difference and a priority, wherein the target historical recommended resource is a historical recommended resource corresponding to a time difference less than a preset time difference threshold.

10. The apparatus of claim 6, wherein, The obtaining unit is further configured to: obtain the plurality of resources corresponding to the target user that are recalled by a plurality of different recalling methods.

11. An electronic device, comprising: comprise: at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-5.

12. A non-transitory computer-readable storage medium having stored thereon computer instructions, wherein, The computer instructions are used to enable the computer to perform the method of any one of claims 1-5.

13. A computer program product, comprising: A computer program, when executed by a processor, implements the method of any one of claims 1-5.

Images