CN119782560B - Multi-genre resource recommendation method and device, electronic equipment and storage medium - Google Patents
Multi-genre resource recommendation method and device, electronic equipment and storage mediumInfo
- Publication number
- CN119782560B CN119782560B CN202411731162.2A CN202411731162A CN119782560B CN 119782560 B CN119782560 B CN 119782560B CN 202411731162 A CN202411731162 A CN 202411731162A CN 119782560 B CN119782560 B CN 119782560B
- Authority
- CN
- China
- Prior art keywords
- resource
- genre
- factor
- resources
- acquiring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The disclosure provides a multi-genre resource recommendation method, a multi-genre resource recommendation device, electronic equipment and a storage medium, and relates to the technical fields of resource distribution, resource recommendation, artificial intelligence, large models and the like. The method comprises the steps of obtaining factor values of at least two fusion factors of each genre resource of at least two genres, obtaining factor parameters of each fusion factor based on a pre-learned strategy model, obtaining a comprehensive factor value of each corresponding genre resource based on the factor values and the factor parameters, and recommending the resource to a user based on the comprehensive factor values. The technology disclosed by the invention can provide an effective implementation scheme for recommending the multi-genre resources in the immersive recommendation scene, and can effectively ensure the recommendation efficiency of the multi-genre resource recommendation.
Description
Technical Field
The disclosure relates to the technical field of computers, in particular to the technical fields of resource distribution, resource recommendation, artificial intelligence, large models and the like, and particularly relates to a multi-genre resource recommendation method, a multi-genre resource recommendation device, electronic equipment and a storage medium.
Background
Based on the increasing demands for consumption efficiency and popularity of immersive consumption habits, more and more users enjoy immersive consumption of resources.
In the existing immersive recommendation scene, the method is mainly used for recommending video resources. The user enters the immersive recommendation scene through the video portal. In the immersive recommendation scene, users do not need to search independently, only need to slide up and down, the immersive recommendation system can continuously recommend videos to the users all the time, and the users can consume the videos immersively. In addition, in the existing immersive recommendation scene, the resources can be ordered through content warmth and the like of the resources so as to realize resource recommendation.
Disclosure of Invention
The disclosure provides a multi-genre resource recommendation method and device, electronic equipment and storage medium.
According to an aspect of the present disclosure, there is provided a multi-genre resource recommendation method, including:
obtaining factor values of at least two fusion factors of each genre resource of at least two genres;
Acquiring factor parameters of each fusion factor based on a pre-learned strategy model;
Based on the factor values and the factor parameters, acquiring the comprehensive factor values of the corresponding genre resources;
And recommending resources to the user based on the comprehensive factor value.
According to another aspect of the present disclosure, there is provided a recommendation device for multi-genre resources, including:
The factor acquisition module is used for acquiring factor values of at least two fusion factors of each genre resource of at least two genres;
The parameter acquisition module is used for acquiring factor parameters of each fusion factor based on a pre-learned strategy model;
The comprehensive value acquisition module is used for acquiring the comprehensive factor value of each corresponding genre resource based on the factor value and the factor parameter;
And the recommending module is used for recommending resources to the user based on the comprehensive factor value.
According to still another aspect of the present disclosure, there is provided an electronic apparatus including:
at least one processor, and
A memory communicatively coupled to the at least one processor, wherein,
The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the aspects and methods of any one of the possible implementations described above.
According to yet another aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the method of the aspects and any possible implementation described above.
According to yet another aspect of the present disclosure, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the method of the aspects and any one of the possible implementations described above.
According to the technology disclosed by the invention, an effective scheme can be provided for recommending the multi-genre resources in the immersive recommendation scene, and the recommendation efficiency of the multi-genre resource recommendation can be effectively ensured.
It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the disclosure, nor is it intended to be used to limit the scope of the disclosure. Other features of the present disclosure will become apparent from the following specification.
Drawings
The drawings are for a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:
FIG. 1 is a schematic diagram according to a first embodiment of the present disclosure;
FIG. 2 is a schematic diagram according to a second embodiment of the present disclosure;
FIG. 3 is a schematic diagram according to a third embodiment of the present disclosure;
FIG. 4 is a schematic diagram according to a fourth embodiment of the present disclosure;
FIG. 5 is a schematic diagram according to a fifth embodiment of the present disclosure;
Fig. 6 is a block diagram of an electronic device for implementing the methods of embodiments of the present disclosure.
Detailed Description
Exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present disclosure to facilitate understanding, and should be considered as merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
It will be apparent that the described embodiments are some, but not all, of the embodiments of the present disclosure. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments in this disclosure without inventive faculty, are intended to be within the scope of this disclosure.
It should be noted that, the terminal device according to the embodiments of the present disclosure may include, but is not limited to, a smart device such as a mobile phone, a Personal digital assistant (Personal DIGITAL ASSISTANT, PDA), a wireless handheld device, a Tablet Computer (Tablet Computer), and the like, and the display device may include, but is not limited to, a device with a display function such as a Personal Computer, a television, and the like.
In addition, the term "and/or" is merely an association relation describing the association object, and means that three kinds of relations may exist, for example, a and/or B, and that three kinds of cases where a exists alone, while a and B exist alone, exist alone. In addition, the character "/" herein generally indicates that the front and rear associated objects are an "or" relationship.
Fig. 1 is a schematic diagram of a first embodiment of the disclosure, and as shown in fig. 1, the embodiment provides a multi-genre resource recommendation method, which specifically includes the following steps:
S101, obtaining factor values of at least two fusion factors of each genre resource of at least two genres;
The execution subject of the multi-genre resource recommendation method in this embodiment may be a multi-genre resource recommendation device, which may be an electronic entity or may also be a software integrated application.
The multi-genre resource recommendation method can be applied to an immersive resource recommendation scene. Moreover, in the immersive resource recommendation scenario of the present embodiment, resources of a plurality of genres may be recommended. The genre of the resource in the present embodiment may refer to a style of the resource, or a format of the resource.
The immersive resource recommendation in this embodiment specifically refers to actively recommending a plurality of resources to a user without engagement without any request by the user. After consuming one of the resources, the user directly slides down and can continue to consume the next resource.
For example, the various genres in the present embodiment may include video resources and teletext resources. Further, the video resources may further include short video resources and/or small video resources, where the playing duration of the small video is less than the playing duration of the short video. Further, the teletext resources may also comprise dynamic teletext resources and/or text teletext resources. The dynamic graphics must include a picture, and may also include a small amount of text to describe the picture. While text teletext must comprise text, possibly a small number of pictures, for interpreting the text.
In this embodiment, for each resource of each genre, a factor value of at least two corresponding fusion factors needs to be obtained, where each fusion factor may be considered as a feature corresponding to the resource. In order to be able to take into account the value of each resource in the following, at least two fusion factors, or at least two features, need to be obtained for each resource in the present implementation.
S102, acquiring factor parameters of each fusion factor based on a pre-learned strategy model;
The method comprises the steps of obtaining factor parameters of the same fusion factor based on a policy model for each genre type of resource, wherein the factor parameters of the same fusion factor obtained for different genre types of resource are different. For example, the policy model of the present embodiment may be an evolution policy model, and may learn by using a large amount of data on-line, so as to evolve to an optimal state. When the step is used on line, the parameters which have evolved to be optimal can be obtained, and then factor parameters of fusion factors of resources of each genre type can be accurately obtained.
S103, acquiring a comprehensive factor value of each corresponding genre resource based on the factor value and the factor parameter;
For example, for each resource, based on the obtained factor values of at least two fusion factors of the resource and the factor parameters of each fusion factor, the comprehensive factor value of the resource can be obtained through a mathematical calculation mode.
And S104, recommending the resources to the user based on the comprehensive factor value.
In this embodiment, the integrated factor value of each resource may be used to comprehensively characterize the recommended degree of the resource. For example, the higher the composite factor value, the higher the recommendation level, i.e., the more worth the resource is recommended.
According to the multi-genre resource recommendation method, the comprehensive factor value of each resource in the plurality of genre resources is obtained based on the factor values of at least two fusion factors of the resources and the factor parameters of each fusion factor, so that various genre resources can be uniformly recommended to a user according to the comprehensive factor values, an effective implementation scheme can be provided for recommending the multi-genre resources in an immersive recommendation scene, and the recommendation efficiency of the multi-genre resource recommendation can be effectively ensured. Moreover, by adopting the technical scheme of the embodiment, the accuracy of resource sequencing can be effectively improved by accurately acquiring the comprehensive factor value of each genre resource, and the efficiency of resource recommendation can be effectively improved.
Fig. 2 is a schematic diagram of a second embodiment of the disclosure, and a multi-genre resource recommendation method according to the present embodiment, based on the technical solution of the embodiment shown in fig. 1, further describes the technical solution of the disclosure in more detail. As shown in fig. 2, the method for recommending multi-genre resources in this embodiment may specifically include the following steps:
S201, acquiring target ordering parameters and first estimated effect characteristics of each genre resource in the corresponding genre type;
For example, in the resource recommendation scenario of the practical application, before the technical solution of the embodiment, the resources of various genres are recalled at the recall layer. And then the coarse and fine arranging layers are processed, wherein the coarse and fine arranging layers can specifically comprise a coarse arranging layer and a fine arranging layer. The method can be used for respectively carrying out coarse sorting and fine sorting on the plurality of resources of the genre type in the resources of various genre types, so as to screen out a plurality of resources with better quality. According to the technical scheme, after coarse and fine arrangement, a plurality of resources of various genre types are placed together for comprehensive sorting, and resource recommendation can be further performed based on comprehensive sorting results.
In this embodiment, the target ranking parameter of each resource in the corresponding genre type resource may be considered to be obtained based on the ranking parameter of each resource in the fine ranking layer in the corresponding genre type resource. The ranking parameter may include a ranking score.
Optionally, in this embodiment, the at least two fusion factors may include a target ranking parameter of each resource within the resources of the corresponding genre type, and at least one first predicted effect feature. The first estimated effect characteristic refers to the effect characteristic of the predicted user when consuming the resource. The user consumes resources, namely, browsing or watching resources by a user, such as browsing video resources or graphic resources by the user.
For example, the first estimated effect feature of each resource in this embodiment may include at least one of a predicted consumption duration, a quick-slip probability, an interaction probability, a play-out probability, a quality score, a slip-down rate, and a drop-out rate of each genre resource.
The consumption time length represents the consumption time length of the user for the resource, the longer the consumption time length is, the higher the satisfaction degree of the user for the resource is, the more the resource is worth recommending to the user, and vice versa.
The faster the probability of a user's consumption of the resource, the less satisfied the user is with the resource, the less worth the resource is recommended to the user, and vice versa.
The interaction in this embodiment may include at least one of praise, forwarding, focusing, sharing, and comment. When a user consumes a resource, the larger the interaction probability is, the higher the attention degree of the user to the resource is, the more worth recommending the resource to the user is, and vice versa.
The probability of completing the playing represents the probability of complete playing when the user consumes the resource, and the larger the probability of completing the playing represents the higher the satisfaction degree of the user on the resource, the more worth the resource is recommended to the user, and vice versa.
The quality score may be considered as a score that comprehensively characterizes the quality of the resource, with higher quality scores indicating better quality of the resource, more worth recommending to the user, and vice versa. Specifically, the quality score may be obtained by evaluating using a quality evaluation strategy or a quality evaluation model.
The slip-down rate indicates the probability that the user will slip down again to consume the next resource after clicking the current resource. The higher the slip-down rate, the higher the user's satisfaction with the current resource, the more worth the resource is recommended to the user, and vice versa.
The drop-out rate represents the probability of dropping out when the user consumes the current resource. The higher the drop-out rate, the higher the user's dissatisfaction with the current resource, the less worth the resource is recommended to the user, and vice versa.
In practical applications, other pre-estimated effect features, such as pre-estimated delay value, may also be included. The estimated post-delay value is used for representing whether the user can continue to watch other resources of an author of the resource or whether the user can continue to watch other related resources of the resource, such as related content resources or related label resources, after consuming the resource.
In this embodiment, each first pre-estimated effect feature may be pre-trained with a corresponding pre-estimated model, and when in use, at least one of a resource feature of a resource, a user feature of a user who wants to consume the resource, a feature of a scene, an inlet genre feature of the resource, and the like may be input into the pre-estimated model, where the pre-estimated model may predict and output the corresponding pre-estimated effect feature based on the input information. Each feature may be represented by a value, where the fast-sliding probability, the interaction probability, the playing probability, the quality score, the sliding down rate, the exiting rate, etc., may be values between 0 and 1, and a higher value may represent a greater probability of the corresponding effect.
In this embodiment, at least two fusion factors of each resource are formed by acquiring the target ordering parameter of each resource in the resource corresponding to the genre type and at least one of the consumption time, the quick slip probability, the interaction probability, the complete broadcast probability, the quality score, the slip-down rate and the withdrawal rate of each resource, so that the acquired fusion factors can be effectively ensured to be quite reasonable and accurate, and effective support is provided for the subsequent calculation of the comprehensive factor value of each resource.
For example, in this embodiment, obtaining the ranking score of each genre resource in the plurality of genre resources of the plurality of genres within the resource of the corresponding genre type may include the following two cases:
The first case, the target ranking parameter, includes the original ranking parameter of the corresponding genre resource in the fine ranking layer within the corresponding genre resource type.
According to the method, the original ordering parameters of the resources in the fine ordering layer in the corresponding genre type resources are adopted as the target ordering parameters of the resources, so that the accuracy of the target ordering parameters of the resources can be effectively ensured.
In case two, not only the ordering parameter of each resource in the corresponding genre type resource is considered, but also the satisfaction degree of each resource is considered, for example, the method specifically comprises the following steps:
(1) Acquiring original ordering parameters of the genre resources in the fine-ranking layer in the corresponding genre types;
(2) Based on a pre-trained satisfaction estimation model, acquiring satisfaction of each genre resource;
(3) Based on satisfaction and original sorting parameters, obtaining corresponding target sorting parameters;
For example, in this embodiment, for each genre resource, the product of the original ranking parameter of the genre resource located in the fine ranking layer within the resource of the corresponding genre type and the satisfaction degree of the resource may be taken as the target ranking parameter of the resource. According to the method, the ranking parameters of the resources are updated by adopting the satisfaction degree of the resources, so that the target ranking parameters of the resources are more reasonable and accurate.
S202, normalizing factor values of the appointed type according to the standard of the corresponding resource genre type, wherein the fusion factors of the appointed type comprise fusion factors with different resource consideration standards of different genre types;
in a practical application scenario, the consideration criteria of factor values of the same fusion factor for resources of different genre types may be different, or may be said to be different in dimension. For example, for the same fusion factor f1, the value range of the fusion factor in the video genre is 0-150, and in the image-text genre, the value range of the fusion factor is 0-100, and the standards of the fusion factor and the image-text genre are different, if the fusion factor f1 and the image-text genre are different, the recommendation of resources of various genre types is not fair and accurate when the comprehensive factor value of the resources is calculated according to the respective values. Therefore, in order to solve this problem, in this embodiment, the normalization processing may be performed on the resources of different genre types and the factor values of the fusion factors with different consideration criteria according to the criteria of the genres of the corresponding resources.
For example, for a fusion factor f1 of a video genre, values in the range of 0-150 are normalized to between 0-1. For the fusion factor f1 of the image-text genre, values within the range of 0-100 are normalized to 0-1. Therefore, the same fusion factor f1 is unified with the standards of different genre resources, so that the standards of the same fusion factor of the resources of different genres are the same, and further the comprehensive factor value of calculating various genre resources can be obtained fairly, reasonably and accurately, and the resource recommendation of multiple genres is realized fairly, reasonably and accurately.
S203, acquiring factor parameters of fusion factors of various genre resources based on a pre-learned evolution strategy (Evolution Strategies; ES) model;
In this embodiment, a covariance matrix adaptive (Covariance Matrix Adaptation; CMA) ES algorithm learned in advance may be specifically adopted to obtain factor parameters of each fusion factor of various genre resources.
Specifically, in at least two fusion factors of each genre resource, factor parameters of the same fusion factor are the same. And in at least two fusion factors of different genre resources, factor parameters of the same fusion factor are different.
For factor parameters of at least two fusion factors of each genre resource, the evolution strategy model may be pre-learned offline based on historical traffic. In the learning process, multiple groups of parameters of at least two fusion factors can be configured for each genre resource, then flow is allocated to each group of parameters, and the effect of each group of parameters is detected, for example, for various parameters, the total consumption time brought by the group of parameters can be adopted as the effect of the group of parameters. And then evolving the parameters of each group so that the parameters of each group are adjusted towards a better effect. Through iterative learning of the round, multiple groups of parameters are evolved, and the optimal parameters of at least two fusion factors can be obtained and used as factor parameters.
Optionally, in one embodiment of the present disclosure, the parameters of each fusion factor of the various genre resources may include a factor index of each fusion factor.
Further alternatively, in case the genre resource comprises a teletext resource, the factor parameter of the fusion factor may also comprise a bias value.
That is, this step S203 may include the step of obtaining an index of each fusion factor of various genre resources based on the evolution strategy model. The exponent here refers to the power. For example, f1 may be f1 a, f1 may be the fusion factor, a may be the index of the fusion factor, or f1 ζ may be used to represent the relationship.
Further, when the resource genre includes the image-text resource, the bias value of each fusion factor of the image-text resource can be obtained based on the evolution strategy model.
In particular, in the conventional immersive resource recommendation scenario, resources for video genre are mainly recommended. In the immersive resource recommendation scenario of the embodiment, the immersive resource recommendation scenario may be used to recommend resources of multiple genres, for example, resources of video genres and resources of image-text genres may be recommended. Compared with the resources of video genres, the resources of the image-text genres are newly added genres in the embodiment, so that the resources of different genres can be placed together, and the video genres are more objectively, equitably and accurately ordered. Specifically, the bias value of each fusion factor of the graphic resource can also be learned in advance through an evolution strategy model.
In addition, in one embodiment of the present disclosure, if the resources of the multiple genres include the resources of the text-to-text genre and the resources of the dynamic-to-text genre, the resources of the text-to-text genre and the resources of the dynamic-to-text genre need to be learned in advance through the evolution policy model, respectively.
S204, for each resource in a plurality of resources of a plurality of genres, acquiring a comprehensive factor value of the resource based on the factor value of each fusion factor corresponding to the resource and the factor parameter of each fusion factor;
Specifically, for a resource of a video genre, a comprehensive factor value of the resource may be obtained according to factor values of at least two fusion factors of the resource and an index of each fusion factor.
For example, taking an example that a resource includes f1, f2, and f3, and taking an example that the index of the fusion factor f1 is a1, the index of the fusion factor f2 is a2, and the index of the fusion factor f3 is a3, the comprehensive score of the resource can be expressed by the following formula:
S=f1a1*f2a2*f3a3
Specifically, for the resource of the graphic genre, the comprehensive factor value of the resource can be obtained according to the factor values of at least two fusion factors of the resource, the index of each fusion factor and the offset value of each fusion factor.
For example, taking an example that one resource includes f1, f2, and f3, taking an example that the index of the fusion factor f1 is a1, the index of the fusion factor f2 is a2, the index of the fusion factor f3 is a3, the offset value of the fusion factor f1 is b1, the index of the fusion factor f2 is b2, and the index of the fusion factor f3 is b3, the comprehensive score of the resource can be expressed by the following formula:
S=(f1-b1)a1*(f2-b2)a2*(f3-b3)a3
S205, sorting a plurality of resources of a plurality of genres according to the comprehensive factor value;
S206, based on the ranking of the comprehensive factor values of the plurality of resources of the plurality of genres, N resources with the top ranking are obtained, and immersive resource recommendation is performed to the user.
Steps S205-S206 of this embodiment are a specific implementation manner of step S104 of the embodiment shown in fig. 1. The number of N may be set based on the requirements of the immersive resource recommendation scenario, for example, may be 6 or 8, which is not limited herein.
According to the multi-genre resource recommendation method, by adopting the technical scheme, the factor values of the fusion factors with different genres and different consideration standards can be normalized in the resources with different genres, so that the standards of the values of the fusion factors of the resources with different genres are unified, and further the subsequent multi-genre resource recommendation can be more reasonably, fairly, accurately and effectively performed. Moreover, by adopting the technical scheme, the comprehensive factor value of each resource can be accurately and efficiently calculated, and further, the recommendation of the multi-genre resources can be more accurately performed based on the sequencing of the comprehensive factor values of each resource. According to the technical scheme, the accuracy of resource sequencing can be further effectively improved, and further the efficiency of resource recommendation can be effectively improved.
Fig. 3 is a schematic diagram of a third embodiment of the disclosure, and a multi-genre resource recommendation method according to the present embodiment further describes the technical scheme of the disclosure in more detail on the basis of the technical scheme of the embodiment shown in fig. 2. As shown in fig. 3, this embodiment provides a specific implementation manner of "predicting satisfaction of each resource based on a pre-trained satisfaction estimation model" in the embodiment shown in fig. 2, which may specifically include the following steps:
s301, obtaining user characteristics of a user;
in this embodiment, the acquisition and use of the user features is known and agreed upon by the user.
For example, this step, when embodied, may include obtaining at least one of an attribute characteristic of the user and a historical consumption characteristic of the user;
The attribute characteristics of the user include at least one of basic attribute characteristics of the user and preference characteristics of the user, and the basic attribute characteristics of the user include age, sex, occupation type, and the like of the user. The preference characteristics of the user include interests of the user, etc. The user's preference characteristics may be identified in the user's attribute information in the form of tags, for example, interests may include travel, entertainment, football, and the like. The attribute characteristics of the user have certain relevance with the resources, and can provide reference for the satisfaction degree prediction of the user on the resources. For example, different ages may have different resource preferences, such as middle and old ages like to consume video resources, middle and old aged office workers like to consume graphics and text resources, middle and old ages like to consume fashion resources, middle and old ages like to consume financial resources, and old ages like to consume health care resources, etc.
The historical consumption characteristics of the user include at least one of a consumption ratio of the user to consume the resources of the various genres, a duration of consuming the resources of the various genres, and a rate of completion of the resources of the various genres for a plurality of historical time periods prior to the current time period. If the currently predicted genre of the resource belongs to the genre of the resource with higher consumption proportion in a plurality of historical time periods, the corresponding satisfaction degree is higher. Conversely, if the currently predicted genre of the resource belongs to the genre of the resource with the lowest consumption ratio in a plurality of historical time periods, the corresponding satisfaction degree is lower. Similarly, the genre of the currently predicted resource belongs to the resource with higher playing rate in a plurality of historical time periods, and the corresponding satisfaction degree is higher. Otherwise, if the genre of the currently predicted resource belongs to the resource with a lower completion rate in a plurality of historical time periods, the corresponding satisfaction degree is lower.
The length of the current period may be a preset time length defined in the resource recommendation scenario, e.g., the current period may be the current day, the current week, etc. The plurality of historical time periods may be configured according to requirements, and may include, for example, one historical day, three historical days, seven historical days, etc., and the length and number of the specific historical time periods are not limited herein. The consumption proportion of the resources of various genres consumed by the user in each historical time period, the duration of consuming the resources of various genres and the rate of completing the playing of the resources of various genres can be obtained through statistics of the historical consumption information of the user.
Further alternatively, in this embodiment, scene features may be acquired, where the scene features can identify that the current immersive resource recommendation scene may be represented as belonging to a recommendation scene or as finding a scene. The recommended scene and the found scene of the embodiment may be two different sections in the application. Different users may have different scene preferences based on which the accuracy of predicted user satisfaction with the resource can be improved.
S302, for each resource, acquiring the resource characteristics of the resource;
For example, the resource characteristics of the resource may include at least one of a resource length, a resource identification, a resource tag, a resource genre, a resource entry genre, and a second pre-estimated effectiveness characteristic of the resource;
When the resource genre type is a video resource, the resource length refers to the duration of the video, and when the resource genre type is an image-text resource, the resource length refers to the length of the text included in the resource and the number of the included pictures;
the resource entry genre feature of the embodiment may refer to a genre type feature of an entry resource of a resource, and can identify that the genre type of the resource entry resource is video or graphics;
the second predicted effect feature of the resource of the embodiment may include at least one of a predicted consumption time period, a fast-sliding probability, an interaction probability, a complete-playing probability, a quality score, a slip-down rate, and a drop-out rate of the resource.
Specifically, the second predicted effect feature of the resource may further include a predicted post-delay value, where the predicted post-delay value is used to characterize whether the predicted user will continue to watch other resources of the author of the resource or whether the predicted user will continue to watch other related resources of the resource, such as related content resources or related label resources, after consuming the resource.
S303, for each resource, based on the user characteristics of the user and the resource characteristics of the resource, adopting a pre-trained satisfaction degree estimation model to predict the satisfaction degree of the user on the resource, and taking the satisfaction degree of the user on the resource as the satisfaction degree of the corresponding resource.
In specific prediction, for each resource, the resource characteristics of the resource and the user characteristics of the user are input into a satisfaction degree prediction model, and the satisfaction degree prediction model can predict and output a satisfaction degree. In practical applications, the satisfaction degree may be a value between 0 and 1, and the larger the value, the higher the satisfaction degree of the resource, and the more worth being recommended.
Optionally, if the scene features are obtained at the same time, the scene features, the features of the resources and the user features are input into the satisfaction degree prediction model together, so that the accuracy of satisfaction degree predicted by the satisfaction degree prediction model is improved.
The satisfaction degree estimating method of the resources can adopt a satisfaction degree estimating model to estimate the satisfaction degree of the user on the resources efficiently and accurately, and provides effective support for resource recommendation. Further, the accuracy of resource sequencing can be further effectively improved through the accuracy of the resource satisfaction degree obtained by the embodiment, and further the efficiency of resource recommendation can be effectively improved.
In addition, in the embodiment, the user characteristics of the user and the resource characteristics of the resource adopted in the satisfaction degree estimation are very rich, and the accuracy of the satisfaction degree estimation can be effectively improved.
Fig. 4 is a schematic diagram of a fourth embodiment of the disclosure, where a multi-genre resource recommendation device 400 is provided, and the recommendation device includes:
A factor obtaining module 401, configured to obtain factor values of at least two fusion factors of each genre resource of at least two genres;
A parameter obtaining module 402, configured to obtain factor parameters of each fusion factor based on a pre-learned policy model;
a comprehensive value obtaining module 403, configured to obtain a comprehensive factor value of each corresponding genre resource based on the factor value and the factor parameter;
and a recommending module 404, configured to recommend resources to the user based on the integrated factor value.
The implementation principle and the technical effect of the multi-genre resource recommendation device 400 in this embodiment are the same as those of the related method embodiments, and detailed descriptions of the related method embodiments may be referred to herein and are not repeated.
Fig. 5 is a schematic diagram of a fifth embodiment of the disclosure, and a multi-genre resource recommendation device 500 according to the present embodiment, based on the technical solution of the embodiment shown in fig. 4, further describes the technical solution of the disclosure in more detail. As shown in fig. 5, the multi-genre resource recommendation device 500 of the present embodiment includes the same name and function modules shown in fig. 4, a factor obtaining module 501, a parameter obtaining module 502, an integrated value obtaining module 503, and a recommendation module 504.
Wherein, factor acquisition module 501 is used for:
acquiring target ordering parameters and first estimated effect characteristics of each genre resource in the corresponding genre type;
The first estimated effect feature comprises at least one of predicted consumption duration, quick slip probability, interaction probability, complete play probability, quality score, slip-down rate and withdrawal rate of each genre resource.
Further optionally, in an embodiment of the present disclosure, the factor obtaining module 501 is configured to:
acquiring original ordering parameters of the genre resources in the fine-ranking layer in the corresponding genre types;
based on a pre-trained satisfaction degree estimation model, acquiring satisfaction degree of each genre resource;
and acquiring corresponding target sorting parameters based on the satisfaction and the original sorting parameters.
Further optionally, in an embodiment of the present disclosure, the factor obtaining module 501 is configured to:
acquiring user characteristics of the user;
Acquiring resource characteristics of each genre resource;
And based on the user characteristics and the resource characteristics, acquiring the corresponding satisfaction degree of the genre resource by adopting the satisfaction degree estimation model.
Further optionally, in one embodiment of the present disclosure, the user characteristics include at least one of an attribute characteristic and a historical consumption characteristic;
The attribute features include at least one of a base attribute feature and a preference feature, and the historical consumption feature includes at least one of a consumption ratio of resources of each genre consumed by the user in at least two historical time periods before the current period, a duration of consumption of resources of each genre, and a rate of completion of the playing of the resources of each genre.
Further optionally, in one embodiment of the present disclosure, the resource characteristics include at least one of a resource length, a resource identification, a resource label, a resource genre, a resource entry genre, and a second pre-estimated effect characteristic;
in the case that the resource genre type includes a video type, the resource length includes a duration of the video;
in case the resource genre type comprises a teletext type, the resource length comprises the length of the text in the teletext and the number of pictures.
Further alternatively, as shown in fig. 5, in an embodiment, the multi-genre resource recommendation device 500 further includes:
the normalization module 505 is configured to normalize the factor value of the specified type according to a standard of a corresponding resource genre, where the fusion factor of the specified type includes fusion factors with different resource consideration standards of different genre types.
Further optionally, in an embodiment of the present disclosure, the factor parameter comprises a factor index.
Further optionally, in an embodiment of the present disclosure, where the genre resource comprises a teletext resource, the factor parameter comprises a bias value.
The implementation principle and the technical effect of the multi-genre resource recommendation device 500 in this embodiment are the same as those of the related method embodiments, and detailed descriptions of the related method embodiments may be referred to herein and are not repeated.
In the technical scheme of the disclosure, the acquisition, storage, application and the like of the related user personal information all conform to the regulations of related laws and regulations, and the public sequence is not violated.
According to embodiments of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.
Fig. 6 illustrates a schematic block diagram of an example electronic device 600 that may be used to implement embodiments of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the disclosure described and/or claimed herein.
As shown in fig. 6, the apparatus 600 includes a computing unit 601 that 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 device 600 may also be stored. The computing unit 601, ROM 602, and RAM 603 are connected to each other by a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.
Various components in the device 600 are connected to the I/O interface 605, including an input unit 606, e.g., keyboard, mouse, etc., an output unit 607, e.g., various types of displays, speakers, etc., a storage unit 608, e.g., magnetic disk, optical disk, etc., and a communication unit 609, e.g., network card, modem, wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.
The computing unit 601 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 601 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as the methods described above of the present disclosure. For example, in some embodiments, the above-described methods of the present disclosure may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 600 via the ROM 602 and/or the communication unit 609. When the computer program is loaded into RAM 603 and executed by computing unit 601, one or more steps of the above-described methods of the present disclosure described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the above-described methods of the present disclosure in any other suitable manner (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), complex Programmable Logic Devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include being implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be a special or general purpose programmable processor, operable 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.
Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may 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 a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
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 may also be used to provide for interaction with a user, for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback), and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background 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 background, 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.
The computer system may include a client and a server. The client and server are typically 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 may be a cloud server, a server of a distributed system, or a server incorporating a blockchain.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present disclosure may be performed in parallel or sequentially or in a different order, provided that the desired results of the technical solutions of the present disclosure are achieved, and are not limited herein.
The above detailed description should not be taken as limiting the scope of the present disclosure. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present disclosure are intended to be included within the scope of the present disclosure.
Claims (15)
1. A multi-genre resource recommendation method, comprising:
obtaining factor values of at least two fusion factors of each genre resource of at least two genres;
Acquiring factor parameters of each fusion factor based on a pre-learned strategy model, wherein the factor parameters comprise factor indexes, and when the genre resources comprise image-text resources, the factor parameters comprise bias values;
Based on the factor values and the factor parameters, acquiring the comprehensive factor values of the corresponding genre resources;
Based on the comprehensive factor value, recommending resources to a user;
the obtaining the factor values of at least two fusion factors of each genre resource of at least two genres comprises:
acquiring target ordering parameters and first estimated effect characteristics of each genre resource in the corresponding genre type;
The first estimated effect characteristics comprise at least one of predicted consumption time length, quick slip probability, interaction probability, playing probability, quality score, sliding down rate and withdrawal rate of each genre resource;
the obtaining the target sorting parameters of the genre resources in the corresponding genre types includes:
acquiring original ordering parameters of the genre resources in the fine-ranking layer in the corresponding genre types;
based on a pre-trained satisfaction degree estimation model, acquiring satisfaction degree of each genre resource;
and acquiring corresponding target sorting parameters based on the satisfaction and the original sorting parameters.
2. The method of claim 1, wherein the target ranking parameter is further capable of being replaced with an original ranking parameter of the corresponding genre resource within the corresponding genre resource type in a fine ranking layer.
3. The method of claim 1, wherein the obtaining satisfaction of each of the genre resources based on the pre-trained satisfaction estimation model comprises:
acquiring user characteristics of the user;
Acquiring resource characteristics of each genre resource;
And based on the user characteristics and the resource characteristics, acquiring the corresponding satisfaction degree of the genre resource by adopting the satisfaction degree estimation model.
4. The method of claim 3, wherein the user characteristics include at least one of attribute characteristics and historical consumption characteristics;
The attribute features include at least one of a base attribute feature and a preference feature, and the historical consumption feature includes at least one of a consumption ratio of resources of each genre consumed by the user in at least two historical time periods before the current period, a duration of consumption of resources of each genre, and a rate of completion of the playing of the resources of each genre.
5. The method of claim 3, wherein the resource characteristics include at least one of a resource length, a resource identification, a resource label, a resource genre, a resource entry genre, and a second pre-estimated effectiveness characteristic;
in the case that the resource genre type includes a video type, the resource length includes a duration of the video;
in case the resource genre type comprises a teletext type, the resource length comprises the length of the text in the teletext and the number of pictures.
6. The method of claim 1, wherein after the obtaining factor values of at least two fusion factors for each of at least two genres, and before obtaining a composite factor value for each of the corresponding genre resources based on the factor values and the factor parameters, the method further comprises:
And normalizing the factor value of the appointed type according to the standard of the corresponding resource genre type, wherein the fusion factors of the appointed type comprise fusion factors with different resource consideration standards of different genre types.
7. A multi-genre resource recommendation device, comprising:
The factor acquisition module is used for acquiring factor values of at least two fusion factors of each genre resource of at least two genres;
The parameter acquisition module is used for acquiring factor parameters of each fusion factor based on a pre-learned strategy model, wherein the factor parameters comprise factor indexes;
The comprehensive value acquisition module is used for acquiring the comprehensive factor value of each corresponding genre resource based on the factor value and the factor parameter;
The recommending module is used for recommending resources to a user based on the comprehensive factor value;
the factor obtaining module is used for:
acquiring target ordering parameters and first estimated effect characteristics of each genre resource in the corresponding genre type;
The first estimated effect characteristics comprise at least one of predicted consumption time length, quick slip probability, interaction probability, playing probability, quality score, sliding down rate and withdrawal rate of each genre resource;
the factor obtaining module is used for:
acquiring original ordering parameters of the genre resources in the fine-ranking layer in the corresponding genre types;
based on a pre-trained satisfaction degree estimation model, acquiring satisfaction degree of each genre resource;
and acquiring corresponding target sorting parameters based on the satisfaction and the original sorting parameters.
8. The apparatus of claim 7, wherein the target ranking parameter is further capable of being replaced with an original ranking parameter of the corresponding genre resources within the corresponding genre resource type in a fine ranking layer.
9. The apparatus of claim 7, wherein the factor acquisition module is configured to:
acquiring user characteristics of the user;
Acquiring resource characteristics of each genre resource;
And based on the user characteristics and the resource characteristics, acquiring the corresponding satisfaction degree of the genre resource by adopting the satisfaction degree estimation model.
10. The apparatus of claim 9, wherein the user characteristics include at least one of attribute characteristics and historical consumption characteristics;
The attribute features include at least one of a base attribute feature and a preference feature, and the historical consumption feature includes at least one of a consumption ratio of resources of each genre consumed by the user in at least two historical time periods before the current period, a duration of consumption of resources of each genre, and a rate of completion of the playing of the resources of each genre.
11. The apparatus of claim 9, wherein the resource characteristics include at least one of a resource length, a resource identification, a resource label, a resource genre type, a resource entry genre, and a second pre-estimated effectiveness characteristic;
in the case that the resource genre type includes a video type, the resource length includes a duration of the video;
in case the resource genre type comprises a teletext type, the resource length comprises the length of the text in the teletext and the number of pictures.
12. The apparatus of claim 7, wherein the apparatus further comprises:
And the normalization module is used for normalizing the factor value of the appointed type according to the standard of the corresponding resource genre type, wherein the fusion factors of the appointed type comprise fusion factors with different resource consideration standards of different genre types.
13. An electronic device, comprising:
at least one processor, and
A memory communicatively coupled to the at least one processor, wherein,
The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-6.
14. A non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the method of any one of claims 1-6.
15. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411731162.2A CN119782560B (en) | 2024-11-28 | 2024-11-28 | Multi-genre resource recommendation method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411731162.2A CN119782560B (en) | 2024-11-28 | 2024-11-28 | Multi-genre resource recommendation method and device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN119782560A CN119782560A (en) | 2025-04-08 |
| CN119782560B true CN119782560B (en) | 2025-09-16 |
Family
ID=95242877
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202411731162.2A Active CN119782560B (en) | 2024-11-28 | 2024-11-28 | Multi-genre resource recommendation method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN119782560B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112163159A (en) * | 2020-10-09 | 2021-01-01 | 北京百度网讯科技有限公司 | Resource recommendation and parameter determination method, device, equipment and medium |
| WO2024169596A1 (en) * | 2023-02-15 | 2024-08-22 | 北京字跳网络技术有限公司 | Multimedia resource processing method and apparatus, device and storage medium |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109408714B (en) * | 2018-10-15 | 2021-02-05 | 上海趣蕴网络科技有限公司 | Multi-model fusion recommendation system and method |
| CN113918764B (en) * | 2020-12-31 | 2024-06-25 | 浙江大学 | A movie recommendation system based on cross-modal fusion |
| CN113254679B (en) * | 2021-06-28 | 2021-09-28 | 北京达佳互联信息技术有限公司 | Multimedia resource recommendation method and device, electronic equipment and storage medium |
| CN116450940A (en) * | 2023-04-12 | 2023-07-18 | 百度在线网络技术（北京）有限公司 | Resource recommendation and estimation model acquisition method, device, electronic equipment and medium |
| CN116578725A (en) * | 2023-05-15 | 2023-08-11 | 北京字跳网络技术有限公司 | A search result sorting method, device, computer equipment and storage medium |
| CN117131264A (en) * | 2023-08-09 | 2023-11-28 | 北京百度网讯科技有限公司 | Resource recommendation methods, devices, electronic equipment and storage media |
-
2024
- 2024-11-28 CN CN202411731162.2A patent/CN119782560B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112163159A (en) * | 2020-10-09 | 2021-01-01 | 北京百度网讯科技有限公司 | Resource recommendation and parameter determination method, device, equipment and medium |
| WO2024169596A1 (en) * | 2023-02-15 | 2024-08-22 | 北京字跳网络技术有限公司 | Multimedia resource processing method and apparatus, device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN119782560A (en) | 2025-04-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114169418B (en) | Tag recommendation model training method and device, tag acquisition method and device | |
| WO2017071251A1 (en) | Information pushing method and device | |
| US20170270455A1 (en) | Managing A Workflow Of Human Intelligence Tasks Based On Task Performance | |
| WO2020238502A1 (en) | Article recommendation method and apparatus, electronic device and storage medium | |
| CN113360761A (en) | Information flow recommendation method and device, electronic equipment and computer-readable storage medium | |
| CN109993627A (en) | Recommended method, the training method of recommended models, device and storage medium | |
| CN112905885A (en) | Method, apparatus, device, medium, and program product for recommending resources to a user | |
| CN113360756A (en) | Resource recommendation method and device, electronic equipment and medium | |
| CN114154013A (en) | Video recommendation method, device, equipment and storage medium | |
| CN115994263A (en) | Resource recommendation method, resource recommendation model training method and device | |
| CN112241327A (en) | Shared information processing method and device, storage medium and electronic equipment | |
| US20160019400A1 (en) | Recommendation system for protecting user privacy | |
| CN114880599B (en) | Information ordering method and device, electronic equipment and storage medium | |
| CN112632384B (en) | Data processing method, device, electronic device and medium for application program | |
| CN119782560B (en) | Multi-genre resource recommendation method and device, electronic equipment and storage medium | |
| CN113269600B (en) | A method and device for sending information | |
| CN119293324A (en) | Content recommendation data processing method, device, equipment and medium based on large model | |
| CN118555455A (en) | Method and device for training sorting model and sorting | |
| CN114328994B (en) | Multimedia resource recommendation method, device, server, storage medium and product | |
| CN119338549B (en) | Training method, device, equipment and storage medium of genre resource allocation model | |
| CN119741078B (en) | Satisfaction degree estimation model training method, device, equipment and storage medium | |
| CN119338547B (en) | Method, device, equipment and storage medium for recommending entry resources | |
| CN117421471A (en) | Diversity-aware model training method, recommendation sorting method and device | |
| CN119338548A (en) | Resource recommendation method, device, electronic device and storage medium | |
| CN120723977B (en) | Resource recommendation methods, model training methods, devices, equipment, and media |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |