Multimedia object recommendation method and device, equipment, storage medium and program product
The multimedia object recommendation method improves accuracy and simplifies user interaction by predicting revisit duration and probability, addressing inefficiencies in existing systems and enhancing user experience.
Patent Information
- Application Number
- CN202411267123.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-10
- Publication Date
- 2024-12-10
AI Technical Summary
Existing multimedia object recommendation systems lack accuracy and efficiency in predicting user revisit behavior, leading to suboptimal user experience and increased operational complexity.
A multimedia object recommendation method that predicts revisit duration and probability for candidate objects, using machine learning models to determine and recommend target multimedia objects based on these parameters, thereby improving accuracy and simplifying user interaction.
Enhances user satisfaction and convenience by accurately predicting and recommending multimedia objects, reducing search time and improving user experience through targeted recommendations.
Smart Images
Figure CN119106194A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of computer technology, specifically to the field of data processing related technology, and more particularly to a multimedia object recommendation method, a multimedia object recommendation device, an electronic device, a computer-readable storage medium, and a computer program product. Background Art
[0002] With the rapid development of information technology, we can analyze users' historical browsing of various types of information, recommend content that users may be interested in, and display the mined content that users may be interested in to users to enhance user satisfaction. Summary of the invention
[0003] The embodiments of the present disclosure provide a multimedia object recommendation method, device, electronic device, computer-readable storage medium, and computer program product. The multimedia object recommendation solution provided by the embodiments of the present disclosure can improve the accuracy of multimedia object recommendation, further simplify the user's operation process for the target multimedia object, enhance user satisfaction, and improve the convenience of users' return visits to the target multimedia object, thereby improving user experience.
[0004] In a first aspect, an embodiment of the present disclosure proposes a multimedia object recommendation method, including: determining a target account's revisit prediction parameter for a candidate multimedia object that has been visited; wherein the revisit prediction parameter includes a predicted revisit duration for the target account to repeatedly visit the candidate multimedia object; determining a target multimedia object from the candidate multimedia objects based on the revisit prediction parameter; and recommending the target multimedia object to the target account.
[0005] In a second aspect, the disclosed embodiment proposes a multimedia object recommendation device, comprising: a first determination module, a second determination module and a push module. The first determination module is configured to determine a return visit prediction parameter of a target account for a candidate multimedia object that has been visited; the return visit prediction parameter includes a return visit prediction duration of the target account repeatedly visiting the candidate multimedia object; the second determination module is configured to determine a target multimedia object from the candidate multimedia objects according to the return visit prediction parameter; and the push module is configured to recommend the target multimedia object to the target account.
[0006] In a third aspect, an embodiment of the present disclosure provides an electronic device, comprising: at least one processor; and a memory communicatively connected to 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 so that the at least one processor can implement the multimedia object recommendation method described in any of the above implementation methods when executing.
[0007] In a tenth aspect, an embodiment of the present disclosure provides a non-transitory computer-readable storage medium storing computer instructions, which are used to enable a computer to implement the multimedia object recommendation method described in any implementation method of the first to third aspects when executed.
[0008] In an eleventh aspect, an embodiment of the present disclosure provides a computer program product including a computer program, which, when executed by a processor, can implement the multimedia object recommendation method described in any one of the implementations of the first to third aspects.
[0009] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] Other features, objects and advantages of the present disclosure will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:
[0011] Figure 1 is an exemplary system architecture in which the present disclosure may be applied;
[0012] Figure 2 A flowchart of a multimedia object recommendation method provided by an embodiment of the present disclosure;
[0013] Figure 3 A flowchart of another multimedia object recommendation method provided by an embodiment of the present disclosure;
[0014] Figure 4 A framework diagram of a model network structure in an application scenario provided by an embodiment of the present disclosure;
[0015] Figure 5 A structural block diagram of a multimedia object recommendation device provided by an embodiment of the present disclosure;
[0016] Figure 6 A schematic diagram of the structure of an electronic device suitable for executing a multimedia object recommendation method provided by an embodiment of the present disclosure. DETAILED DESCRIPTION
[0017] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description. It should be noted that the embodiments in the present disclosure and the features in the embodiments may be combined with each other without conflict.
[0018] It should be pointed out that in the technical solution of the present invention, the collection, acquisition, storage, processing, transmission, provision, disclosure and application of user personal information (such as relevant information records of user access to multimedia objects, etc.) are all carried out with the user's knowledge and explicit authorization, comply with the relevant laws and regulations, and do not violate public order and good morals.
[0019] Figure 1 An exemplary system architecture 100 is shown to which embodiments of the multimedia object recommendation scheme of the present disclosure may be applied.
[0020] like Figure 1 As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.
[0021] The user can use the terminal devices 101, 102, 103 to interact with the server 105 through the network 104 to receive or send messages, etc. The terminal devices 101, 102, 103 and the server 105 can be installed with various applications for realizing information communication between the two, such as multimedia object recommendation applications, audio object recommendation applications, video object recommendation applications, instant messaging applications, etc.
[0022] Terminal devices 101, 102, 103 and server 105 can be hardware or software. When terminal devices 101, 102, 103 are hardware, they can be various electronic devices with display screens, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, etc.; when terminal devices 101, 102, 103 are software, they can be installed in the electronic devices listed above, which can be implemented as multiple software or software modules, or as a single software or software module, and no specific limitation is made here. When server 105 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, or as a single software or software module, and no specific limitation is made here.
[0023] The server 105 can provide various services through various built-in applications. For example, the server 105 can provide multimedia object recommendation related applications. When running the multimedia object recommendation related applications, the server 105 can achieve the following effects: by predicting the revisit prediction parameters that characterize the revisit or repeated visit of the target account to the candidate multimedia objects that have been visited, at least the predicted revisit duration of the target account's repeated visits to the candidate multimedia objects can be predicted. Then, the target multimedia objects that are ultimately recommended to the target account can be screened out from the candidate multimedia objects according to the revisit prediction parameters, which can improve the accuracy of multimedia object recommendations, and further simplify the user's operating procedures for the target multimedia objects, enhance user satisfaction, and improve the convenience of users' revisiting the target multimedia objects, thereby improving user experience.
[0024] It should be noted that, in addition to being obtained from the terminal devices 101, 102, and 103 through the network 104, the relevant information of the user's historical access to the multimedia object can also be pre-stored in the server 105 in various ways. Therefore, when the server 105 detects that the data has been stored locally (for example, the relevant information of the user's historical access to the multimedia object will be regularly backed up or stored in the server 105 before starting the processing), the server 105 can choose to directly obtain the data locally. In this case, the exemplary system architecture 100 may also not include the terminal devices 101, 102, 103 and the network 104.
[0025] Since it takes up more computing resources and stronger computing power to predict the relevant information of the multimedia objects visited by the user in history in order to recommend the multimedia objects, the multimedia object recommendation method provided in the subsequent embodiments of the present disclosure is generally executed by the server 105 with stronger computing power and more computing resources, and accordingly, the multimedia object recommendation device is generally also set in the server 105. However, it should also be pointed out that when the terminal devices 101, 102, and 103 also have computing power and computing resources that meet the requirements, the terminal devices 101, 102, and 103 can also complete the above-mentioned operations that are handed over to the server 105 through the multimedia object recommendation application installed thereon, and then output the same result as the server 105. Especially in the case where there are multiple terminal devices with different computing powers at the same time, but when the multimedia object recommendation application determines that the terminal device where it is located has stronger computing power and more remaining computing resources, it can let the terminal device perform the above-mentioned operations, thereby appropriately reducing the computing pressure of the server 105, and accordingly, the multimedia object recommendation device can also be set in the terminal devices 101, 102, and 103. In this case, the exemplary system architecture 100 may also not include the server 105 and the network 104 .
[0026] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.
[0027] Please refer to Figure 2 , Figure 2 This is a flowchart of a multimedia object recommendation method provided by an embodiment of the present disclosure. The multimedia object recommendation method is executed by an execution subject server, wherein process 200 includes the following steps:
[0028] Step 201: Determine a return visit prediction parameter of a target account for a candidate multimedia object that the target account has visited; wherein the return visit prediction parameter includes a return visit prediction duration of the target account repeatedly visiting the candidate multimedia object.
[0029] This step is intended to be performed by the above-mentioned execution subject of the multimedia object recommendation method (for example Figure 1 The server 105 shown in the figure determines the return visit prediction parameter of the target account for the candidate multimedia object that has been visited, wherein the return visit prediction parameter may at least include the return visit prediction duration of the target account repeatedly visiting the candidate multimedia object; wherein the target account is used to indicate a specific user, and the target account can be represented by a user identifier (ID), or a combination of a user identifier and an account login password. It should be noted that the return visit prediction parameter is determined with the knowledge and consent of the user corresponding to the target account.
[0030] In some optional implementations of the embodiments of the present disclosure, the candidate multimedia objects visited by the target account may include at least one of the following: multimedia objects visited by the target account for the first time but not completed; multimedia objects visited by the target account multiple times but not completed; multimedia objects visited by the target account for the first time and completed; multimedia objects visited by the target account multiple times and completed. Among them, the standard for completing the access to the multimedia object may include playing the entire duration of the multimedia object or playing to the end of the multimedia object (such as the ending song and / or the ending easter egg) but playing its entire duration.
[0031] In some optional implementations of the embodiments of the present disclosure, the candidate multimedia objects may include audio objects and / or video objects. The audio object may be one or more of an audio book, a radio play, an audio book, etc.; the video object may be a long video whose duration is different from a short video, such as a TV series, a movie, etc.
[0032] In some optional implementations of the embodiments of the present disclosure, the target account repeatedly accessing the candidate multimedia object may be the user continuing or accessing (such as listening to or watching) a multimedia object that has been accessed but not completed. In some other optional implementations of the embodiments of the present disclosure, the target account repeatedly accessing the candidate multimedia object may be the target account re-accessing a multimedia object that has been accessed.
[0033] In some optional implementations of the embodiments of the present disclosure, the above-mentioned return visit prediction parameter can represent the prediction result of the target account's return visit or repeated visit to the candidate multimedia object after the target account has visited the candidate multimedia object for the first time (for example, within a certain period of time after the current visit). Among them, the above-mentioned return visit prediction duration can represent the estimated duration of the target account's return visit or repeated visit to the candidate multimedia object after the target account has visited the candidate multimedia object for the first time (for example, within a certain period of time after the current visit). Among them, the number of the above-mentioned candidate multimedia objects and the number of target multimedia objects can be one or more, respectively, depending on the specific situation, and are not specifically limited here.
[0034] Step 202: Determine a target multimedia object from candidate multimedia objects according to the return visit prediction parameter.
[0035] This step is intended to be performed by the above-mentioned execution subject of the multimedia object recommendation method (for example Figure 1The server 105 shown in the figure), based on the above step 201, after determining the above return prediction parameters for the above candidate multimedia objects visited by the target account, the multimedia objects to be recommended (i.e., the target multimedia objects) can be screened from the candidate multimedia objects based on the return prediction parameters.
[0036] Step 203: Recommend the target multimedia object to the target account.
[0037] According to the multimedia object recommendation method provided by the embodiment of the present disclosure, by predicting the revisit prediction parameters that characterize the revisit or repeated visit of the target account to the candidate multimedia objects that have been visited, at least the prediction of the revisit prediction duration of the target account's repeated visits to the candidate multimedia objects can be specifically realized, and then the target multimedia objects that are finally recommended to the target account can be screened out from the candidate multimedia objects according to the revisit prediction parameters, which can improve the accuracy of multimedia object recommendation and further simplify the user's operation process for the target multimedia object. For example, it can at least save the user's time in searching and locating the target multimedia object, enhance user satisfaction and improve the convenience of the user's revisit to the target multimedia object, thereby improving the user experience.
[0038] In the above Figure 2 On the basis of the corresponding embodiment, in response to the above-mentioned return visit prediction parameter including the return visit prediction duration of the target account repeatedly visiting the candidate multimedia object, and implementing accurate multimedia object recommendation according to the return visit prediction duration, the following implementation methods may be specifically adopted but are not limited to:
[0039] In a specific implementation of the embodiment of the present disclosure, the candidate multimedia objects may be first sorted in descending order according to their respective corresponding predicted return times, and then the candidate multimedia objects corresponding to the predicted return times at the first preset position may be directly determined as the target multimedia objects, wherein the value of the first preset position may be set according to the specific circumstances and is not specifically limited herein; or the candidate multimedia objects whose corresponding predicted return times are greater than or equal to the first specified time may be directly determined as the target multimedia objects, wherein the value of the first specified time may be set according to the specific circumstances and is not specifically limited herein; or the candidate multimedia objects whose predicted return times are greater than or equal to the second specified time may be first screened out, and then the predicted return times corresponding to these candidate multimedia objects may be sorted in descending order, and then the candidate multimedia objects corresponding to the predicted return times at the second preset position may be determined as the target multimedia objects, wherein the values of the second specified time and the values of the second preset position may be set according to the specific circumstances and are not specifically limited herein. Among them, the value of the above-mentioned first preset position and the value of the above-mentioned second preset position can be the same or different, and the value of the above-mentioned first specified time length and the value of the second specified time length can be the same or different, and no specific limitation is made here.
[0040] In another specific implementation of the embodiment of the present disclosure, the initial recommendation score of the candidate multimedia object can be obtained first, and then the initial recommendation score is multiplied by the corresponding normalized (for example, processed to a value between 0 and 1) predicted return time, and the result of the product operation is used as the recommendation score of the candidate multimedia object; then the candidate multimedia objects can be sorted in descending order according to their corresponding recommendation scores, and then the candidate multimedia object corresponding to the recommendation score ranked in the first third preset position is directly determined as the target multimedia object, wherein the value of the first third preset position can be set according to the specific situation and is not specifically limited here; or the candidate multimedia object can be sorted according to the recommendation score of the candidate multimedia object. The candidate multimedia objects whose recommendation scores are greater than or equal to the first score threshold are directly determined as the target multimedia objects, wherein the size of the first score threshold can be set according to the specific situation and is not specifically limited here; or the candidate multimedia objects whose recommendation scores are greater than or equal to the second score threshold can be first screened out, and then the recommendation scores of these candidate multimedia objects are sorted in order from large to small, and then the candidate multimedia objects corresponding to the recommendation scores ranked in the first fourth preset position are determined as the target multimedia objects, wherein the size of the first score threshold, the size of the second score threshold, and the value of the first fourth preset position can be set according to the specific situation and are not specifically limited here. Among them, the values of the first third preset position and the first fourth preset position can be the same or different, and the value of the first score threshold and the value of the second score threshold can be the same or different, and are not specifically limited here.
[0041] Furthermore, in the above Figure 2 On the basis of the corresponding embodiments, in some optional implementations of the embodiments of the present disclosure, the above-mentioned return visit prediction parameters may include not only the return visit prediction duration of the target account's repeated visits to the above-mentioned candidate multimedia objects, but also the return visit prediction probability of the target account's repeated visits to the candidate multimedia objects. Among them, the return visit prediction probability can characterize the possibility that the target account will return or repeatedly visit the visited candidate multimedia object (for example, within a certain period of time after the current visit) after the current visit to the candidate multimedia object. By further evaluating the return visit prediction probability of the target account's repeated visits when recommending multimedia objects to the target account, the accuracy of multimedia object recommendations can be further improved.
[0042] Further, in response to the above-mentioned return visit prediction parameters including the return visit prediction duration and return visit prediction probability of the target account repeatedly visiting the candidate multimedia object, when accurate multimedia object recommendation is implemented according to the return visit prediction duration and return visit prediction probability, the following implementation methods may be specifically adopted but are not limited to:
[0043] In a specific implementation of the disclosed embodiment, the predicted duration of return visits and the predicted probability of return visits of the candidate multimedia objects, which have been normalized (for example, processed to a value between 0 and 1), can be multiplied to obtain a recommendation score for the candidate multimedia object; and then, based on the recommendation score of the candidate multimedia object, the same part of the target multimedia object recommended to the target account is determined from the candidate multimedia objects. For related content, please refer to the corresponding part of the above-mentioned embodiment of determining the target multimedia object based on the recommendation score of the candidate multimedia object, which will not be repeated here.
[0044] In another specific implementation method of the disclosed embodiment, an initial recommendation score of the candidate multimedia object may be first obtained, and then the initial recommendation score may be multiplied by the corresponding normalized predicted duration of revisit and predicted probability of revisit (e.g., processed to a value between 0 and 1) to obtain the recommendation score of the candidate multimedia object; and then, based on the recommendation score of the candidate multimedia object, the same part of the target multimedia object recommended to the target account may be determined from the candidate multimedia objects. For related content, please refer to the corresponding part of the above-mentioned embodiment of determining the target multimedia object based on the recommendation score of the candidate multimedia object, which will not be repeated here.
[0045] It should be noted that the above-mentioned implementation method of determining the target multimedia object to be recommended to the target account from the candidate multimedia objects based on the predicted duration of the return visit or based on the predicted duration of the return visit and the predicted probability of the return visit is only a partial implementation method of determining the target multimedia object to be recommended to the target account from the candidate multimedia objects based on the return visit prediction parameters. Other implementation methods may also be used in other embodiments of the present disclosure, which are not specifically limited here.
[0046] Please refer to Figure 3 , Figure 3 This is a flowchart of another multimedia object recommendation method provided by an embodiment of the present disclosure. The multimedia object recommendation method is executed by an execution subject server, wherein process 300 includes the following steps:
[0047] Step 301: Determine return visit prediction parameters of a target account corresponding to at least two different preset times for a candidate multimedia object.
[0048] This step is intended to be performed by the above-mentioned execution subject of the multimedia object recommendation method (for example Figure 1 The server 105 shown in the figure determines the return visit prediction parameters corresponding to at least two different preset times for the candidate multimedia objects visited by the target account. For the same content of the return visit prediction parameters, please refer to the corresponding part of the previous embodiment, which will not be repeated here. It should be noted that the return visit prediction parameters are determined with the knowledge and consent of the user corresponding to the target account.
[0049] In some optional implementations of the embodiments of the present disclosure, the interval between any two preset times is based on natural days. In this way, accurate prediction of the return visits or repeated visits of the target account to the multimedia object over a long period of time can be achieved.
[0050] The at least two different preset times may be at least two different times after the target account's last access to the candidate multimedia object, and the values of the at least two different preset times may be set according to specific circumstances, such as setting at least two different preset times corresponding to different types of the candidate multimedia object. Specifically, in some specific implementations of the embodiments of the present disclosure, the at least two different preset times may include at least two of the following day, 7 days, 15 days, and 30 days after the target account's last access to the candidate multimedia object.
[0051] Furthermore, in some optional implementations of the embodiments of the present disclosure, the above step 301 can be specifically executed as follows: obtaining historical data on multimedia object accesses of the target account; inputting the historical data into a pre-trained first prediction model to obtain output revisit prediction parameters corresponding to each preset time; wherein the first prediction model is trained based on a training set constructed with sample historical data related to multimedia object accesses as input and corresponding revisit real parameters as expected outputs, and the revisit real parameters correspond to the preset time.
[0052] In this implementation, a first prediction model with return visit prediction parameters corresponding to at least two different preset time periods as optimization targets can be pre-trained, and accurate parameter prediction can be performed based on the trained prediction model, thereby accurately predicting the return visit or repeated visit of the target account to the multimedia object over a long period of time. The first prediction model can be trained based on a deep neural network (DNN) model of a customized gate control network (Customized GateControl, CGC) as a model to be trained, and a training set constructed with sample historical data related to multimedia object access as input and corresponding return visit real parameters as expected output.
[0053] In some optional implementations of the present disclosure, the historical data of the target account's access to the above-mentioned multimedia objects (or the target account's historical data related to multimedia access, or the target account's historical data of accessing multimedia objects) may include but is not limited to user-side feature data, resource-side feature data and request-side feature data; wherein the user-side feature data may include but is not limited to the number of times the target account accesses (such as clicks) each multimedia object every day within a period of time (such as the last 7 days or 30 days, etc.), the duration of each visit, the time, and the target account's identifier (userid), gender, age, occupation, historical consumption records, etc.; the resource-side feature data may include but is not limited to the identifier of the multimedia object (resource id), author information, category, etc.; the request-side feature data request predicted time, device type, etc.
[0054] For details, please refer to Figure 4 The model network structure shown in FIG. 1 , the sample history data related to the multimedia object access can be input through the embedding layer, and the CGC network can at least include an expert network structure and a prediction tower, wherein the expert network structure responsible for learning can be divided into all task sharing (such as Figure 4The shared expert module shown in ) and task-specific expert (such as Figure 4 The expert modules 1-6 shown in FIG. 1 are related to two categories, wherein each prediction tower (such as Figure 4 The prediction towers 1-6 shown in FIG. 1 correspond to specific return visit prediction parameters (such as Figure 4 The 1-day revisit rate and revisit time, 7-day revisit rate and revisit time, and 30-day revisit rate and revisit time shown in are predicted for output. Among them, the 1-day revisit rate (re1) can refer to the probability of whether the corresponding account will listen to or watch the same multimedia object (such as the same audio book) again within the next day after listening or watching it that time; the 1-day revisit time (re1pl) can refer to the total time that the corresponding account will listen to or watch the same multimedia object again within the next day after listening or watching it that time; the 7-day revisit rate (re7) can refer to whether the corresponding account will listen to or watch the same multimedia again within the next 7 days after listening or watching it that time. object; 7-day revisit time (re7pl) may refer to the total time that the corresponding account listens to or watches the same multimedia object again within the next 7 days after the current listening or viewing; 30-day revisit rate (re30) may refer to whether the corresponding account listens to or watches the same multimedia object again within the next 30 days after the current listening or viewing; 30-day revisit time (re30pl) may refer to the total time that the corresponding account listens to or watches the same multimedia object again within the next 30 days after the current listening or viewing.
[0055] Step 302: Determine the recommendation score of the candidate multimedia object according to the return visit prediction parameter.
[0056] Furthermore, in some optional implementations of the embodiments of the present disclosure, step 302 may be specifically performed as follows: obtaining an initial recommendation score for the candidate multimedia object; obtaining the result of multiplying the initial recommendation score and the revisit prediction parameters corresponding to each preset time; and determining the result as the recommendation score.
[0057] In this embodiment, an initial recommendation score of a candidate multimedia object can be first obtained, and then the initial recommendation score can be multiplied by a return visit prediction parameter corresponding to each preset time in at least two different preset times that have been normalized (for example, processed to a value between 0 and 1), and the result of the operation can be used as the recommendation score of the candidate multimedia object, thereby accurately screening out the target multimedia object to be recommended to the user from the candidate multimedia objects to improve user satisfaction.
[0058] Further, in an optional specific implementation method of the embodiment of the present disclosure, in response to the return prediction parameters of the candidate multimedia object including a return prediction duration and a return prediction probability, the at least two different preset times include preset time 1 to preset time n, where n is an integer greater than or equal to 2, then the recommendation score of the candidate multimedia object = the candidate's initial recommendation score for the media object * the return prediction duration corresponding to preset time 1 * the return prediction probability corresponding to preset time 1 * … * the return prediction duration corresponding to preset time n * the return prediction probability corresponding to preset time n.
[0059] Step 303: Determine the target multimedia object from the candidate multimedia objects according to the recommendation score, so that the target multimedia object can be accurately recommended.
[0060] Further, in some optional implementations of the embodiments of the present disclosure, the target multimedia object includes: a candidate multimedia object corresponding to a target recommendation score in the recommendation score that is greater than a preset score value.
[0061] In this embodiment, after determining the recommendation score of the candidate multimedia objects, the multimedia objects to be recommended to the target account, i.e., the target multimedia objects, can be screened out according to the size relationship between each recommendation score and the preset score value. Specifically, the candidate multimedia objects with a larger value than the preset score can be pushed to the target account first to improve the convenience of users' revisiting the multimedia objects, thereby improving the user experience.
[0062] Among them, the process of taking the candidate multimedia object corresponding to the target recommendation score greater than the preset score value in the recommendation score as the target recommendation object can be understood as directly comparing each recommendation score with the above-mentioned preset score value, screening out the target recommendation score greater than the preset score value, and then the candidate multimedia object corresponding to the target recommendation score can be determined as the target multimedia recommendation object.
[0063] Furthermore, the recommendation scope of the above-mentioned multimedia objects can be further narrowed, that is, according to the relationship between the recommendation score of each candidate multimedia object and the preset score value, first screen out a part of the candidate multimedia objects whose recommendation score is greater than the preset score value, and then sort the part of the candidate multimedia objects according to their corresponding recommendation scores in order from large to small, and then determine the recommendation scores ranked in the top N positions as the above-mentioned target recommendation scores, and then determine the candidate multimedia objects corresponding to the target recommendation scores as the target multimedia objects, wherein the value of N can be set to a positive integer depending on the specific situation, which is not specifically limited here, and the size of the above-mentioned preset score value can be set according to the specific situation, for example, it can be the same as one or more of the recommendation scores of the candidate multimedia objects, or it can be a score value that is different from the recommendation scores of all the candidate multimedia objects, which is not specifically limited here.
[0064] According to the multimedia object recommendation method provided by the embodiment of the present disclosure, by predicting the revisit prediction parameters that characterize the revisit situations or repeated visits of the target account to the candidate multimedia objects visited at at least two different preset times, the target multimedia object that is finally recommended to the target account can be screened out from the candidate multimedia objects according to the revisit prediction parameters corresponding to the at least two different preset times. This can improve the accuracy of multimedia object recommendation and further simplify the user's operation process for the target multimedia object. For example, it can at least save the user's time in searching and locating the target multimedia object, enhance user satisfaction and improve the convenience of the user's revisit to the target multimedia object, thereby improving the user experience.
[0065] In some optional implementations of the multimedia object recommendation method in any of the above embodiments of the present disclosure, the above-mentioned acquisition of the initial recommendation score of the candidate multimedia object includes: inputting the historical data of the multimedia object access of the target account into a pre-trained second prediction model to obtain the initial recommendation parameters of the output candidate multimedia object; wherein the second prediction model is trained based on a training set constructed with sample historical data related to the multimedia object access as input and corresponding real recommendation parameters as expected output; and determining the initial recommendation score based on the initial recommendation parameters.
[0066] In this implementation, a second prediction model with the initial recommendation parameters of multimedia objects as the optimization target can be pre-trained, and accurate parameter prediction can be performed based on the trained prediction model, thereby accurately predicting the access situation (including first access and / or repeated access situation) of the target account to the multimedia object within a relatively short period of time (such as minutes, and further not more than half a day or a day). The second prediction model can be trained with a DNN model as the model to be trained, and a training set constructed with sample historical data related to multimedia object access as input and corresponding real recommendation parameters as expected output.
[0067] In some optional implementations of the present disclosure, the historical data of the target account's access to the above-mentioned multimedia objects (or the historical data of the target account related to the above-mentioned multimedia access, or the historical data of the target account's access to multimedia objects) may include but is not limited to user-side feature data, resource-side feature data and request-side feature data; wherein the user-side feature data may include but is not limited to the number of times the target account has historically accessed (such as clicked) each multimedia object, the duration of each access, and the target account's identifier (user id), gender, age, occupation, historical consumption records, etc.; the resource-side feature data may include but is not limited to the identifier of the multimedia object (resource id), author information, category, etc.; the request-side feature data includes the predicted time of the request, device type, etc.
[0068] It should be noted that the historical data and sample historical data of the corresponding account accessing the multimedia object are obtained with the knowledge and consent of the corresponding user. And the historical data and sample historical data can be data spliced at hourly granularity to facilitate data transmission and statistical analysis of playback time, sample annotation, etc. Specifically, the corresponding historical data and sample historical data can be obtained from the corresponding recommendation log.
[0069] In some optional implementations of the embodiments of the present disclosure, the above-mentioned initial recommendation parameters include the following but are not limited to at least one of: Click Through Rate (CTR), start rate, play duration, and play rate of play duration exceeding a fixed duration. Among them, the start rate may refer to the probability that the target account clicks into (including directly entering and entering by clicking a jump link, etc.) the real or actual play page of the multimedia object and clicks to play, and the fixed duration may be set according to the specific situation. In some implementations, it may be set at a granularity of minutes (such as 5 minutes), which is not specifically limited here. In this way, by predicting the single access behavior of the target account, it is possible to accurately predict the access situation of the target account to the multimedia object within a relatively short period of time (which may include the first visit and / or repeated visit situation).
[0070] In some optional implementations of the embodiments of the present disclosure, the above-mentioned initial recommendation parameters may be parameters related to the target account's first visit to the candidate multimedia object, so as to achieve an accurate prediction of the possibility of a return visit of the target account based on the prediction result of the first multimedia object recommendation to the target account.
[0071] Further references Figure 5 As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of a multimedia object recommendation device. Figure 2-3 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices (such as Figure 1 Server 105 shown in FIG.
[0072] like Figure 5 As shown, the multimedia object recommendation device 500 of the embodiment of the present disclosure may include: a first determination module 501 , a second determination module 502 and a push module 503 .
[0073] Among them, the above-mentioned first determination module 501 is configured to be configured to determine the return visit prediction parameters of the target account for the candidate multimedia objects that have been visited; the return visit prediction parameters include the return visit prediction duration of the target account's repeated visits to the candidate multimedia objects; the above-mentioned second determination module 502 is configured to determine the target multimedia object from the candidate multimedia objects based on the return visit prediction parameters; the above-mentioned push module 503 is configured to recommend the target multimedia object to the target account.
[0074] In the multimedia object recommendation device 500, the specific processing of the first determination module 501, the second determination module 502 and the push module 503 and the technical effects thereof can be referred to in detail. Figure 2 The relevant descriptions of steps 201 - 203 in the corresponding embodiment are not repeated here.
[0075] In some optional implementations of the embodiments of the present disclosure, the above-mentioned return visit prediction parameter also includes a return visit prediction probability of the target account repeatedly visiting the candidate multimedia object.
[0076] In some optional implementations of the embodiments of the present disclosure, the above-mentioned first determination module 501 is specifically configured to: determine the return visit prediction parameters of the target account corresponding to at least two different preset times for the candidate multimedia object; and the push module 503 specifically includes: a first determination sub-module 5031 and a push sub-module 5032.
[0077] The first determination submodule 5031 is configured to determine the recommendation score of the candidate multimedia object according to the return visit prediction parameter; the push submodule 5032 is configured to determine the target multimedia object from the candidate multimedia objects according to the recommendation score.
[0078] In the multimedia object recommendation device 500 of the present disclosure, the specific processing of the first determination module 501, the first determination submodule 5031 and the push submodule 5032 and the technical effects thereof can be referred to in detail. Figure 3 The relevant descriptions of steps 301 - 303 in the corresponding embodiment are not repeated here.
[0079] In some optional implementations of the embodiments of the present disclosure, the interval between any two preset times is based on natural days.
[0080] In some optional implementations of the embodiments of the present disclosure, the target multimedia object includes: a candidate multimedia object corresponding to a target recommendation score in the recommendation score that is greater than a preset score value.
[0081] In some optional implementations of the embodiments of the present disclosure, the first determination module 501 specifically includes: an acquisition submodule 5011 and a second determination submodule 5012 .
[0082] Among them, the above-mentioned acquisition submodule 5011 is configured to obtain the historical data of multimedia object access of the target account; the above-mentioned second determination submodule 5012 is configured to input the historical data into a pre-trained first prediction model to obtain the output of revisit prediction parameters corresponding to each preset time; wherein the first prediction model is based on the sample historical data related to the multimedia object access as input and the corresponding revisit real parameters as the training set constructed with the expected output, and the revisit real parameters correspond to the preset time.
[0083] In some optional implementations of the embodiments of the present disclosure, the above-mentioned first determination submodule 5031 is specifically configured to: obtain an initial recommendation score for the candidate multimedia object; obtain the result of multiplying the initial recommendation score and the revisit prediction parameters corresponding to each preset time; and determine the result as the recommendation score.
[0084] In some optional implementations of the embodiments of the present disclosure, the above-mentioned first determination submodule 5031 is further configured to: input the historical data of multimedia object access of the target account into a pre-trained second prediction model to obtain initial recommendation parameters of the output candidate multimedia objects; wherein the second prediction model is trained based on a training set constructed with sample historical data related to multimedia object access as input and corresponding real recommendation parameters as expected output; and determine the initial recommendation score based on the initial recommendation parameters.
[0085] In some optional implementations of the embodiments of the present disclosure, the above-mentioned initial recommendation parameters include at least one of the following: click-through rate, start rate, play duration, and play rate of play duration exceeding a fixed duration.
[0086] This embodiment exists as an apparatus embodiment corresponding to the above-mentioned method embodiment. By predicting the revisit prediction parameters that characterize the revisit or repeated visit of the target account to the candidate multimedia objects that have been visited, at least the prediction of the revisit prediction duration of the target account's repeated visits to the candidate multimedia objects can be specifically realized. Then, the target multimedia objects that are ultimately recommended to the target account can be screened out from the candidate multimedia objects according to the revisit prediction parameters, which can improve the accuracy of multimedia object recommendations, and further simplify the user's operation process for the target multimedia object, enhance user satisfaction, and improve the convenience of users' revisiting the target multimedia objects, thereby improving user experience.
[0087] Figure 6A schematic block diagram of an example electronic device 600 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.
[0088] like Figure 6 As shown, the electronic device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to 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. In the RAM 603, various programs and data required for the operation of the electronic device 600 can also be stored. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.
[0089] Multiple components in the electronic 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 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 electronic device 600 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.
[0090] The computing unit 601 may be a variety of general and / or special 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 dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 601 performs the various methods and processes described above, such as the multimedia object recommendation method. For example, in some embodiments, the multimedia object recommendation method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the multimedia object recommendation method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the multimedia object recommendation method in any other appropriate manner (e.g., by means of firmware).
[0091] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
[0092] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.
[0093] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0094] To provide interaction with a user, the systems and techniques described herein 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 pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the 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 input, voice input, or tactile input).
[0095] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may 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.
[0096] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, a server of a distributed system, or a server combined with a blockchain. Among them, a cloud server, also known as a cloud computing server or a cloud host, is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and virtual private servers (VPS, Virtual Private Server) services.
[0097] According to the technical solution of the embodiment of the present disclosure, by predicting the revisit prediction parameters that characterize the return visits or repeated visits of the target account to the candidate multimedia objects that have been visited, at least the prediction of the return visit prediction duration of the target account's repeated visits to the candidate multimedia objects can be specifically realized, and then the target multimedia objects that are ultimately recommended to the target account can be screened out from the candidate multimedia objects according to the revisit prediction parameters, which can improve the accuracy of multimedia object recommendations, further simplify the user's operating process for the target multimedia object, enhance user satisfaction, and improve the convenience of users' return visits to the target multimedia objects, thereby improving user experience.
[0098] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.
[0099] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.
Claims
1. A multimedia object recommendation method, comprising: Determine a return visit prediction parameter of the target account for the candidate multimedia object visited; wherein the return visit prediction parameter includes a return visit prediction duration of the target account repeatedly visiting the candidate multimedia object; determining a target multimedia object from the candidate multimedia objects according to the return visit prediction parameter; The target multimedia object is recommended to the target account.
2. The method according to claim 1, wherein: The return visit prediction parameter also includes a return visit prediction probability that the target account repeatedly visits the candidate multimedia object.
3. The method according to claim 1 or 2, wherein: The step of determining a return visit prediction parameter of a target account for a candidate multimedia object that has been visited includes: Determining the return visit prediction parameters of the target account corresponding to at least two different preset times for the candidate multimedia object; and The step of determining the target multimedia object from the candidate multimedia objects according to the return visit prediction parameter comprises: determining a recommendation score of the candidate multimedia object according to the return visit prediction parameter; The target multimedia object is determined from the candidate multimedia objects according to the recommendation score.
4. The method according to claim 3, wherein: The interval between any two preset times is based on natural days.
5. The method according to claim 3, wherein: The target multimedia object includes: a candidate multimedia object corresponding to a target recommendation score greater than a preset score value among the recommendation scores.
6. The method according to claim 3, wherein: The determining the return visit prediction parameters of the target account corresponding to at least two different preset times for the candidate multimedia object includes: Obtaining historical data of multimedia object accesses of the target account; Inputting the historical data into a pre-trained first prediction model to obtain the return visit prediction parameters corresponding to each of the preset times; The first prediction model is obtained by training a training set based on sample historical data related to multimedia object access as input and corresponding real parameters of return visits as expected output, and the real parameters of return visits correspond to the preset time.
7. The method according to claim 3, wherein: The step of determining the recommendation score of the candidate multimedia object according to the return visit prediction parameter comprises: Obtaining an initial recommendation score for the candidate multimedia object; Obtaining a result of multiplying the initial recommendation score and the return visit prediction parameter corresponding to each of the preset time periods; The result is determined as the recommendation score.
8. The method according to claim 7, wherein: The obtaining of the initial recommendation score of the candidate multimedia object comprises: Inputting the historical data of multimedia object access of the target account into a pre-trained second prediction model to obtain the initial recommendation parameters of the candidate multimedia object as output; wherein the second prediction model is trained based on a training set constructed based on sample historical data related to multimedia object access as input and corresponding real recommendation parameters as expected output; The initial recommendation score is determined according to the initial recommendation parameters.
9. The method according to claim 8, wherein: The initial recommended parameters include at least one of the following: Click-through rate, start rate, playback duration, and playback rate that exceeds a fixed duration.
10. A multimedia object recommendation device, comprising: A first determination module is configured to determine a return visit prediction parameter of a target account for a candidate multimedia object that has been visited; wherein the return visit prediction parameter includes a return visit prediction duration of the target account repeatedly visiting the candidate multimedia object; A second determination module is configured to determine a target multimedia object from the candidate multimedia objects according to the return visit prediction parameter; The push module is configured to recommend the target multimedia object to the target account.
11. The device according to claim 10, wherein: The return visit prediction parameter also includes a return visit prediction probability that the target account repeatedly visits the candidate multimedia object.
12. The device according to claim 10 or 11, wherein: The first determination module is specifically configured to: determine the return visit prediction parameters of the target account corresponding to at least two different preset times for the candidate multimedia object; as well as The push module specifically includes: A first determination submodule is configured to determine a recommendation score of the candidate multimedia object according to the return visit prediction parameter; The push submodule is configured to determine the target multimedia object from the candidate multimedia objects according to the recommendation score.
13. The device according to claim 12, wherein: The interval between any two preset times is based on natural days.
14. The device according to claim 12, wherein: The target multimedia object includes: a candidate multimedia object corresponding to a target recommendation score greater than a preset score value among the recommendation scores.
15. The device according to claim 12, wherein: The first determination module specifically includes: An acquisition submodule, configured to acquire historical data of multimedia object accesses of the target account; The second determination submodule is configured to input the historical data into a pre-trained first prediction model to obtain the output return visit prediction parameters corresponding to each of the preset times; wherein the first prediction model is trained based on a training set constructed with sample historical data related to multimedia object access as input and corresponding return visit real parameters as expected output, and the return visit real parameters correspond to the preset time.
16. The device according to claim 12, wherein: The first determining submodule is specifically configured as follows: Obtaining an initial recommendation score for the candidate multimedia object; Obtaining a result of multiplying the initial recommendation score and the return visit prediction parameter corresponding to each of the preset time periods; The result is determined as the recommendation score.
17. The device according to claim 16, wherein: The first determining submodule is further configured as follows: Inputting the historical data of multimedia object access of the target account into a pre-trained second prediction model to obtain the initial recommendation parameters of the candidate multimedia object as output; wherein the second prediction model is trained based on a training set constructed based on sample historical data related to multimedia object access as input and corresponding real recommendation parameters as expected output; The initial recommendation score is determined according to the initial recommendation parameters.
18. The device according to claim 17, wherein: The initial recommended parameters include at least one of the following: Click-through rate, start rate, playback duration, and playback rate that exceeds a fixed duration.
19. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed 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 multimedia object recommendation method according to any one of claims 1 to 9.
20. A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the multimedia object recommendation method according to any one of claims 1 to 9.
21. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method for recommending multimedia objects according to any one of claims 1 to 9 is implemented.