CN115291503B - Recommendation method, recommendation device, electronic device and storage medium - Google Patents
Recommendation method, recommendation device, electronic device and storage medium Download PDFInfo
- Publication number
- CN115291503B CN115291503B CN202211196512.0A CN202211196512A CN115291503B CN 115291503 B CN115291503 B CN 115291503B CN 202211196512 A CN202211196512 A CN 202211196512A CN 115291503 B CN115291503 B CN 115291503B
- Authority
- CN
- China
- Prior art keywords
- recommendation
- media resource
- media
- parameter
- preset
- 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- 230000003993 interaction Effects 0.000 claims abstract description 16
- 230000004044 response Effects 0.000 claims abstract description 9
- 238000012163 sequencing technique Methods 0.000 claims description 16
- 230000008859 change Effects 0.000 claims description 5
- 230000001105 regulatory effect Effects 0.000 abstract description 10
- 230000001276 controlling effect Effects 0.000 abstract description 2
- 238000004590 computer program Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000004891 communication Methods 0.000 description 6
- 230000005540 biological transmission Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 4
- 238000013500 data storage Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 101100481876 Danio rerio pbk gene Proteins 0.000 description 2
- 101100481878 Mus musculus Pbk gene Proteins 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 239000008186 active pharmaceutical agent Substances 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001143 conditioned effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B11/00—Automatic controllers
- G05B11/01—Automatic controllers electric
- G05B11/36—Automatic controllers electric with provision for obtaining particular characteristics, e.g. proportional, integral, differential
- G05B11/42—Automatic controllers electric with provision for obtaining particular characteristics, e.g. proportional, integral, differential for obtaining a characteristic which is both proportional and time-dependent, e.g. P. I., P. I. D.
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Automation & Control Theory (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
The disclosure provides a recommendation method, a recommendation device, an electronic device and a storage medium. A recommendation method may include: in response to receiving a current request of a user account for media resources, obtaining recommendation results of a plurality of candidate media resources by using a recommendation model; for each candidate media resource, under the condition that the candidate media resource is determined to be the media resource of the preset type, obtaining an adjusting parameter for adjusting the recommendation result of the candidate media resource and adjusting the recommendation result by using the adjusting parameter, wherein the interaction times of the media resource of the preset type are less than the preset times; and selecting candidate media resources meeting preset conditions from the plurality of candidate media resources according to the adjusted recommendation result, and using the candidate media resources as the media resources recommended to the user account. The method and the device can ensure that the preset type media resources are recommended more accurately in the recommendation system by regulating and controlling the predicted recommendation result of the preset type media resources (such as cold videos).
Description
Technical Field
The present disclosure relates to the field of data processing, and in particular, to a recommendation method, a recommendation apparatus, an electronic device, and a storage medium based on PID control.
Background
The recommendation system is widely applied to various internet online services, and helps a user to find favorite products, music, videos, books and other media resources more quickly. However, most of the current recommendation systems cause underestimation of the model score of the cold media resource and overestimation of the model score of the hot media resource due to the ubiquitous exposure bias problem, thereby causing a marxiao effect, that is, the hot media resource becomes hotter, and the cold media resource becomes colder, thereby affecting the display and the transmission of new media resources. This results in a tendency for the recommender system to recommend to the user an excessive number of popular media assets, rather than the media assets that are actually of interest to the user.
Accordingly, it is currently sought to provide a recommendation system that reduces the effects of the mataires effect, providing corresponding media assets for user account needs.
This background description is for the purpose of facilitating understanding of relevant art in the field and is not to be construed as an admission of the prior art.
Disclosure of Invention
The present disclosure provides a recommendation method, a recommendation apparatus, an electronic device, and a storage medium to at least solve the above-described problems. The technical scheme of the disclosure is as follows:
according to a first aspect of embodiments of the present disclosure, there is provided a recommendation method, which may include: in response to receiving a current request of a user account for media resources, obtaining recommendation results of a plurality of candidate media resources by using a recommendation model; for each of the candidate media resources, under the condition that the candidate media resource is determined to be a preset type of media resource, obtaining an adjustment parameter for adjusting a recommendation result of the candidate media resource and adjusting the recommendation result by using the adjustment parameter, wherein the interaction times of the preset type of media resource are less than preset times; and selecting candidate media resources meeting preset conditions from the candidate media resources according to the adjusted recommendation result, and using the candidate media resources as the media resources recommended to the user account.
As an embodiment, obtaining an adjustment parameter for adjusting the recommendation result of the candidate media resource may include:
acquiring a proportional parameter, an integral parameter and a differential parameter for adjusting the recommendation result of the candidate media resource; the proportional parameter, the integral parameter and the differential parameter are determined based on a recommendation probability error sequence, and the recommendation probability error is determined based on recommendation probabilities respectively corresponding to the preset type of media resources and the non-preset type of media resources; the proportional parameter represents a recommended probability error closest to the current moment in the recommended probability error sequence, the integral parameter represents an accumulated error of the recommended probability error sequence, and the differential parameter represents a recommended probability error change rate closest to the current moment in the recommended probability error sequence.
As an embodiment, the proportional parameter, the integral parameter and the derivative parameter may be determined by:
for each of at least one preset scheduling period prior to receiving the current request, performing the following: obtaining historical recommendation results of historical media resources in a plurality of historical requests in the preset scheduling period, wherein the historical media resources in the plurality of historical requests comprise the first media resources of the preset type and the second media resources of the non-preset type, and the historical recommendation results are obtained by prediction of the recommendation model; determining a recommendation probability of the recommendation model for the first media resource based on historical recommendations of the first media resource and determining a recommendation probability of the recommendation model for the second media resource based on historical recommendations of the second media resource;
determining the proportional parameter, the integral parameter and the differential parameter according to the recommendation probabilities of the first media resource and the second media resource of the at least one preset scheduling period.
As an embodiment, determining a probability of recommendation of the first media asset by the recommendation model based on the historical recommendation of the first media asset and determining a probability of recommendation of the second media asset by the recommendation model based on the historical recommendation of the second media asset may include:
according to the historical recommendation result of the first media resource, sequencing each of the first media resources in the respective historical request;
calculating the recommendation probability of the recommendation model to the first media resource by comparing the sequencing position of the first media resource with a preset position;
according to the historical recommendation result of the second media resource, sequencing each of the second media resources in the respective historical request;
and calculating the recommendation probability of the recommendation model to the second media resource by comparing the sequencing position of the second media resource with a preset position.
As an embodiment, the determining the proportional parameter, the integral parameter and the derivative parameter according to the recommendation probabilities of the first media resource and the second media resource of the at least one preset scheduling period may include:
calculating a recommendation probability error of each preset scheduling period according to the recommendation probability of the first media resource and the second media resource of each preset scheduling period;
accumulating the recommended probability error of each preset scheduling period to obtain a recommended probability error sequence;
and respectively determining the proportional parameter, the integral parameter and the differential parameter by using the recommended probability error sequence.
As an embodiment, the determining the proportional parameter, the integral parameter and the differential parameter by using the recommended probability error sequence may include:
taking a recommended probability error of a first preset scheduling period closest to the current request as the proportional parameter;
taking the sum of each recommended probability error in the recommended probability error sequence as the integral parameter;
and taking the difference between the recommended probability errors of the first preset scheduling period and the previous preset scheduling period of the first preset scheduling period as the differential parameter.
According to a second aspect of the embodiments of the present disclosure, there is provided a recommendation apparatus, which may include:
a prediction module configured to obtain recommendation results for a plurality of candidate media resources using a recommendation model in response to receiving a current request for a media resource by a user account;
the adjusting module is configured to, for each of the plurality of candidate media resources, obtain an adjusting parameter for adjusting a recommendation result of the candidate media resource and adjust the recommendation result by using the adjusting parameter when the candidate media resource is determined to be a media resource of a preset type, where the number of interactions of the media resource of the preset type is less than a preset number of times;
and the recommending module is configured to select candidate media resources meeting preset conditions from the candidate media resources according to the adjusted recommending result, and the candidate media resources are used as the media resources recommended to the user account.
As an embodiment, the adjustment module may be configured to:
acquiring a proportional parameter, an integral parameter and a differential parameter for adjusting the recommendation result of the candidate media resource; the proportional parameter, the integral parameter and the differential parameter are determined based on a recommendation probability error sequence, and the recommendation probability error is determined based on recommendation probabilities respectively corresponding to the preset type of media resources and the non-preset type of media resources; the proportional parameter represents a recommended probability error closest to the current moment in the recommended probability error sequence, the integral parameter represents an accumulated error of the recommended probability error sequence, and the differential parameter represents a recommended probability error change rate closest to the current moment in the recommended probability error sequence.
As an embodiment, the adjustment module may be configured to determine the proportional parameter, the integral parameter and the derivative parameter by:
for each of at least one preset scheduling period prior to receiving the current request, performing the following: obtaining historical recommendation results of historical media resources in a plurality of historical requests in the preset scheduling period, wherein the historical media resources in the plurality of historical requests comprise the first media resources of the preset type and the second media resources of the non-preset type, and the historical recommendation results are obtained by prediction of the recommendation model; determining a recommendation probability of the recommendation model for the first media resource based on historical recommendations of the first media resource and determining a recommendation probability of the recommendation model for the second media resource based on historical recommendations of the second media resource;
determining the proportional parameter, the integral parameter and the differential parameter according to the recommendation probabilities of the first media resource and the second media resource of the at least one preset scheduling period.
As an embodiment, the adjustment module may be configured to:
according to the historical recommendation result of the first media resource, sequencing each of the first media resources in the respective historical request;
calculating the recommendation probability of the recommendation model to the first media resource by comparing the sequencing position of the first media resource with a preset position;
according to the historical recommendation result of the second media resource, sequencing each of the second media resources in the respective historical request;
and calculating the recommendation probability of the recommendation model to the second media resource by comparing the sequencing position of the second media resource with a preset position.
As an embodiment, the adjustment module may be configured to:
calculating a recommendation probability error of each preset scheduling period according to the recommendation probability of the first media resource and the second media resource of each preset scheduling period;
accumulating the recommended probability error of each preset scheduling period to obtain a recommended probability error sequence;
and respectively determining the proportional parameter, the integral parameter and the differential parameter by using the recommended probability error sequence.
As an embodiment, the adjustment module may be configured to:
taking a recommended probability error of a first preset scheduling period closest to the current request as the proportional parameter;
taking the sum of each recommended probability error in the recommended probability error sequence as the integral parameter;
and taking the difference between the recommended probability errors of the first preset scheduling period and the previous preset scheduling period of the first preset scheduling period as the differential parameter.
According to a third aspect of embodiments of the present disclosure, there is provided an electronic apparatus, which may include: at least one processor; at least one memory storing computer-executable instructions, wherein the computer-executable instructions, when executed by the at least one processor, cause the at least one processor to perform the recommendation method as described above.
According to a fourth aspect of embodiments of the present disclosure, there is provided a computer-readable storage medium storing instructions that, when executed by at least one processor, cause the at least one processor to perform the recommendation method as described above.
According to a fifth aspect of embodiments of the present disclosure, there is provided a computer program product, wherein instructions of the computer program product are executed by at least one processor of an electronic device to perform the recommendation method as described above.
The technical scheme provided by the embodiment of the disclosure at least brings the following beneficial effects:
by regulating and controlling the predicted recommendation result of the preset type media resources (such as cold videos), the preset type media resources can be more accurately recommended in the recommendation system, the Martian effect of the recommendation system is further reduced, the penetration possibility of the preset type media resources is improved, and the media resources which are really interested in the user are provided for the user so as to improve the user experience.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the disclosure and are not to be construed as limiting the disclosure.
FIG. 1 is a flow chart of a recommendation method according to an embodiment of the present disclosure;
FIG. 2 is a flow diagram of a recommendation method according to another embodiment of the present disclosure;
FIG. 3 is a block diagram of a recommendation device according to an embodiment of the present disclosure;
fig. 4 is a schematic structural diagram of a recommendation device according to an embodiment of the present disclosure;
fig. 5 is a block diagram of an electronic device according to an embodiment of the disclosure.
Detailed Description
In order to make the technical solutions of the present disclosure better understood by those of ordinary skill in the art, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.
The following description with reference to the accompanying drawings is provided to assist in a comprehensive understanding of embodiments of the disclosure as defined by the claims and their equivalents. Various specific details are included to aid understanding, but these are to be considered exemplary only. Accordingly, those 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. In addition, descriptions of well-known functions and constructions are omitted for clarity and conciseness.
The terms and words used in the following description and claims are not limited to the written meaning, but are used only by the inventors to achieve a clear and consistent understanding of the disclosure. Accordingly, it should be apparent to those skilled in the art that the following descriptions of the various embodiments of the present disclosure are provided for illustration only and not for the purpose of limiting the disclosure as defined by the appended claims and their equivalents.
It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the above-described drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the disclosure described herein are capable of operation in sequences other than those illustrated or otherwise described herein. The implementations described in the exemplary embodiments below are not intended to represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.
In consideration of the exposure deviation problem existing in the model pre-estimated score of the recommendation system, the method provides a PID control-based fairness regulation algorithm of the recommendation system, namely, based on the idea of PID control, fairness indexes/recommendation probabilities of cold and hot media resources are regulated and controlled, media resources with different enthusiasm are guaranteed to be treated fairly in the recommendation system, the recommendation system can reflect the real interest of users more accurately, and the probability of new video showing is improved.
In the disclosure, considering that the fairness condition of the recommendation system is constantly changing, the estimation score of the recommendation model cannot be regulated and controlled in a fixed parameter manner, and therefore, a common PID algorithm in a cybernetics is used. For example, fairness indexes/recommendation probabilities of cold media resources and hot media resources can be respectively calculated and difference values of the fairness indexes/recommendation probabilities can be obtained; inputting the difference value into a PID control system to obtain a regulation coefficient of the model estimated score of the cold media resource; finally, the model pre-estimated score of the cold media resource is adjusted through the regulation and control coefficient, and the accuracy of the model scoring of the recommendation system can be improved.
In addition, considering that fairness is a group index and an effective index can be obtained only by accumulating certain data, the method of off-line combination is provided for fairness index/recommendation probability regulation and control in the disclosure. For example, small-scale accumulation can be performed on model estimated scores of cold media resources and hot media resources, and a PID regulation coefficient can be calculated off-line; the online service reads the offline PID control coefficient to perform online control according to the specific situation of each flow. Therefore, the timely updating of the regulation and control coefficient can be met on the premise of ensuring that the fairness index/recommendation probability calculation is effective.
Hereinafter, according to various embodiments of the present disclosure, a method and apparatus of the present disclosure will be described in detail with reference to the accompanying drawings.
Fig. 1 is a flow chart of a recommendation method according to an embodiment of the present disclosure. The recommendation method according to embodiments of the present disclosure may be implemented in any electronic device having data processing capabilities. The electronic device may be a device including at least one of, for example, a smart phone, a tablet Personal Computer (PC), a mobile phone, a video phone, an electronic book reader (e-book reader), a desktop PC, a laptop PC, a netbook computer, a workstation, a server, a Personal Digital Assistant (PDA), a Portable Multimedia Player (PMP), a video player, a camera, a wearable apparatus, and a server, etc.
Referring to fig. 1, in step S101, in response to receiving a current request for a media resource from a user account, recommendation results for a plurality of candidate media resources are obtained using a recommendation model. Here, the recommendation results may include a recommendation score for the candidate media resources by the recommendation model. In the fine ranking process of the recommendation system, a trained recommendation model is generally used for further model scoring of the media resources which are coarsely ranked, and then the media resources to be recommended to the user account are screened out according to the recommendation score of the recommendation model. Media assets may refer to video, audio, files, merchandise, etc.
For example, when a user account browses videos in video software, the interaction of the user account with the videos (such as sliding and browsing videos of interest on a touch screen of an electronic device) may be regarded as a user account request, and in this case, the recommendation system may model-score a plurality of candidate videos (such as a video set that is arranged by a recommendation system in a bold line) by using a recommendation model in response to the user account request, and calculate a recommendation result of the plurality of candidate videos.
For another example, when the user account browses goods in the purchasing platform, the interaction of the user account with the goods (such as sliding on a touch screen of the electronic device to browse the goods of interest) may be regarded as a user account request, and in this case, the recommendation system may model-score a plurality of candidate goods (such as the goods that are arranged in a row by the recommendation system) by using the recommendation model in response to the user account request, and calculate recommendation scores of the plurality of candidate goods.
The above examples are merely illustrative, and the present disclosure is not limited thereto.
In step S102, for each of the plurality of candidate media resources, in a case where the candidate media resource is determined as a media resource of a preset type, an adjustment parameter for adjusting a recommendation result of the candidate media resource is acquired and the recommendation result is adjusted using the adjustment parameter. The interaction times of the preset type of media resources are less than the preset times.
Considering that the model estimation score of the recommendation system often has the exposure bias problem, for example, the score of the hot video is higher, and the score of the cold video is lower, which directly results in that the hot video is easier to be revealed and the cold video is difficult to be revealed in the recommendation candidate set finally revealed to the user account. Therefore, the fairness of the cold videos and the hot videos needs to be regulated and controlled, the videos with different enthusiasm are guaranteed to be treated fairly in the recommendation system, the possibility of new video showing is improved, the enthusiasm of the creator is stimulated, and the platform ecology is improved.
In the present disclosure, some type of adjustment that requires model recommendation may be preset. Taking a video as an example, the model recommendation result of the cold video can be preset to be regulated. For example, the types that need model recommendation adjustment may be filtered out according to the number of interactions of the user account with the media resource. When the number of interactions of the candidate media resource is less than the preset number, the candidate media resource may be determined as a preset type of media resource. Taking a video as an example, whether the video belongs to a hot video or a cold video can be judged according to the real showing times of the video. For example, a video may be classified as a popular video when the true number of presentations exceeds 1000. When a video is really presented less than 1000 times, the video may be classified as a cold video.
Considering that the fairness condition of the system is constantly changed and the model pre-estimation score cannot be regulated and controlled in a fixed parameter mode, the PID algorithm is adopted to regulate and control the model pre-estimation score.
For the candidate media resources belonging to the preset type, control parameters for performing proportional, integral and differential regulation on the recommendation results of the candidate media resources can be obtained, and then the recommendation results of the candidate media resources are regulated and controlled by the control parameters.
Since fairness is a group index, certain data accumulation is needed to obtain an effective index. Therefore, off-line integration is proposed in this disclosure for fairness tuning. For example, model scores for cold and hot videos may be accumulated at a certain time granularity (such as an hour scale) and PID regulation coefficients calculated offline; the online request service can read the offline PID coefficient to perform online regulation and control according to the specific condition of each flow. Therefore, timely updating of the regulation and control coefficient can be met on the premise that fairness calculation is guaranteed to be effective.
According to an embodiment of the disclosure, the proportional parameter, the integral parameter, and the differential parameter for adjusting the recommendation result of the candidate media resource may be determined based on a recommendation probability error sequence, and the recommendation probability error may be determined based on recommendation probabilities respectively corresponding to a preset type of media resource and a non-preset type of media resource.
The proportional parameter can represent the recommendation probability error closest to the current moment in the recommendation probability error sequence, the integral parameter can represent the accumulated error of the recommendation probability error sequence, and the differential parameter can represent the recommendation probability error change rate closest to the current moment in the recommendation probability error sequence.
As an example, for each of at least one preset scheduling period before the current request is received, the following operations may be performed: obtaining historical recommendation results of historical media resources in a plurality of historical requests in a preset scheduling period, wherein the historical media resources in the plurality of historical requests comprise a first media resource of a preset type and a second media resource of a non-preset type, and the historical recommendation results are obtained by prediction of a recommendation model; determining a recommendation probability of the recommendation model for the first media resource based on the historical recommendation result of the first media resource and determining a recommendation probability of the recommendation model for the second media resource based on the historical recommendation result of the second media resource; and determining a proportional parameter, an integral parameter and a differential parameter according to the recommendation probability of the first media resource and the second media resource of at least one preset scheduling period. The first media asset and the second media asset may represent different media asset types. Taking video as an example, the first media resource may refer to cold video and the second media resource may refer to hot video. The first media asset and the second media asset may each comprise a plurality of media assets.
For a preset scheduling period, when the recommendation probability is calculated, each of the first media resources and each of the second media resources may be sorted in the respective history requests according to the recommendation results of the first media resources and the second media resources, and the recommendation probability predicted by the recommendation model for the recommendation results of the first media resources and the second media resources is calculated by comparing the sorted positions of the first media resources and the second media resources with the preset position. Specifically, each of the first media resources may be sorted in the respective history request according to the history recommendation result of the first media resource, and the recommendation probability of the recommendation model for the first media resource may be calculated by comparing the sorted position of the first media resource with a preset position. Each of the second media assets can be ranked in its respective history request according to its historical recommendation result; and calculating the recommendation probability of the recommendation model to the second media resource by comparing the sequencing position of the second media resource with a preset position.
For example, the predetermined location may be a kth location of candidate media resources that are coarsely ranked for a history request, candidate media resources ranked before the kth location according to the model are media resources to be recommended to the user account, and candidate media resources ranked after the kth location are media resources not recommended to the user account.
Then, calculating a recommendation probability error of each preset scheduling period according to the recommendation probability of the first media resource and the second media resource of each preset scheduling period, and accumulating the recommendation probability error of each preset scheduling period to obtain a recommendation probability error sequence; and respectively determining a proportional parameter, an integral parameter and a differential parameter for adjusting the recommendation result of the recommendation model by utilizing the recommendation probability error sequence. Specifically, the recommended probability error of the first preset scheduling period closest to the current request may be used as a proportional parameter; taking the sum of each recommended probability error in the recommended probability error sequence as an integral parameter; and taking the difference between the recommended probability errors of the first preset scheduling period and the previous preset scheduling period of the first preset scheduling period as a differential parameter.
The following describes how to determine PID control parameters for cold video in detail by taking video as an example.
For each clicked video (in order to ensure that the video quality is as homogeneous as possible, the clicked video is adopted in the disclosure), the real presentation times of the clicked video can be recorded in real time, and the video higher than a certain threshold (for example, the real presentation times 10000 are used as the threshold) is recorded as a popular videoA video below this threshold is recorded as a cold video ≥>. At the same time, a model score for the video in each request can be obtained, forming<Video, scoring>For the following steps: />、/>Wherein->And &>And respectively scoring models of the hot video and the cold video. According to the model scoring, each video can be sorted in the respective request in a descending order to obtain<Video, ordering>For the following steps: />、/>,/>Or->Smaller means that the corresponding video is ranked higher (i.e., higher scoring) in the respective request. Finally, the fairness index (i.e., the recommendation probability) for cold and hot videos can be calculated according to equations (1) and (2), respectively:
wherein, in the equation (1),N h indicating the number of hot videos,is shown asiThe ranking position of the individual popular videos in their corresponding requests,kindicates a preset position in the corresponding request, and->Represents a decision function when>Is less than or equal tokThen the function outputs 1; when/is>Is greater thankWhen the function outputs 0, the judgment isiWhether a hot video enters a request topk. In the case of the equation (2),N c indicates the number of cold videos, and>is shown asiThe rank position of an individual cold video in its corresponding request,kindicates a preset position in the corresponding request, and->Represents a decision function when>Is less than or equal tokThen the function outputs 1; when/is>Is greater thankWhen the function outputs 0, the judgment isiWhether a hot video enters a request topk. Typically, a preset location in each requestkThe same is true.
The implication of the fairness index/recommendation probability is that the cold/hot video entry request topkThe ratio of (c) to (d), for a fair recommendation system,and/or>Should be very close but in the actual recommendation system->Tend to be greater than>. Thus, the present disclosure can adjust these two recommendation probability values as close as possible through PID regulation.
Based on the above steps of calculating the recommendation probability of the hot/cold video, the recommendation probability error of the cold/hot video may be calculated according to a certain time granularity. For example, the recommended probability error for a cold/hot video may be calculated once per hour according to equation (3):
further obtaining a small-scale recommended probability error sequenceWherein->Indicating the error of the last hour of the day,represents an error from the present farthest time instant>To calculate the time window for the PID control coefficients, the units are hours.
According to the PID control principle, the method can obtain,/>,/>。/>Indicating that the recommended probability error of the first preset scheduling period closest to the current request is taken as a proportional parameter. />The sum of each recommended probability error in the recommended probability error sequence is used as an integral parameter. />The difference between the recommended probability errors of the first preset scheduling period and the scheduling period before the first preset scheduling period is used as a differential parameter.
Thus, for each offline scheduling period (such as per hour), a set of PID coefficients is availableAt the end of each calculation, the coefficients are updated to the cache for on-line reading.
For each online request, a model score of the candidate video may be obtained first to obtain<Video, scoring>To pair. According to the existing real showing times of the current video, whether the video is a cold video can be judged. If the video is a hot video, the model scoring result can not be adjusted, otherwise, PID scoring regulation and control of the cold video are performed. Then, based on the obtained PID coefficient>And super-ginseng>And calculating the score result after regulation as follows:
in step S103, according to the adjusted recommendation result, a candidate media resource meeting a preset condition is selected from the plurality of candidate media resources as a media resource recommended to the user account. For example, for a hot video, the model scoring result can be directly transmitted to the back link sorting module of the recommendation system, and for a cold video, the adjusted model scoring result can be transmitted to the back link sorting module after the model scoring result is regulated. The back link sequencing module sequences the videos according to the scores of all the videos to screen out the topkThe video of (2).
According to the embodiment of the disclosure, a recommendation system fairness regulation and control algorithm based on a PID control idea is provided to ensure that cold videos and hot videos can be treated fairly, the possibility of showing out the cold videos is improved, and the recommendation system can effectively reflect user interests, arouse enthusiasm of creators and improve platform ecology. In addition, any system with unfairness can be regulated and controlled in the above manner.
Fig. 2 is a flow chart of a recommendation method according to another embodiment of the present disclosure. The recommendation method according to embodiments of the present disclosure may be implemented in any electronic device having data processing capabilities. The following describes the recommendation of cold and hot videos by the recommendation system as an example.
Referring to fig. 2, in step S201, an online request for a video by a user account is received. A request for video p by user account u may be represented by < u, p >.
In step S202, for online requests, each candidate video may be model scored using a recommendation model.
In step S203, a model score of each candidate video may be obtained to obtain a score of each candidate video<Video, scoring>To pair。/>Representing a candidate video scored with the corresponding model.
In step S204, it can be determined whether the video is a cold video according to the existing actual showing times of the current video. If the current video is the popular video, the model scoring result is not adjusted, and the step S206 is entered to show the model scoring result of the popular video to the subsequent link module. If the current video is a cold video, step S205 is performed.
In step S205, PID control is performed on the model scoring result of the current video. PID coefficient obtained by off-line calculationAnd super-ginseng>And calculating a model scoring result after regulation:then will->Giving a subsequent link module. The PID control parameters applied to the cold video model scoring of the current online request may be calculated offline according to equations (1) through (3) above.
Fig. 3 is a block diagram of a recommendation device according to an embodiment of the present disclosure.
Referring to fig. 3, the recommendation apparatus 300 may include a prediction module 301, an adjustment module 302, and a recommendation module 303. Each module in the recommendation device 300 may be implemented by one or more modules, and names of the corresponding modules may vary according to types of the modules. In various embodiments, some modules in the recommendation device 300 may be omitted, or additional modules may be included. Furthermore, modules/elements according to various embodiments of the present disclosure may be combined to form a single entity, and thus the functions of the respective modules/elements prior to combination may be equivalently performed.
In response to receiving a current request for a media resource from a user account, the prediction module 301 may obtain recommendations for a plurality of candidate media resources using a recommendation model.
For each of the plurality of candidate media assets, the adjustment module 302 can determine whether the candidate media asset is of a preset type of media asset. In the case that the candidate media resource is determined to be a media resource of the preset type, the adjusting module 302 may obtain an adjusting parameter for adjusting the recommendation result of the candidate media resource and adjust the recommendation result by using the adjusting parameter. The number of interactions of the preset type of media resource may be less than a preset number.
For example, the adjusting module 302 may determine the number of interactions of the candidate media resource, and determine the candidate media resource as a media resource of a preset type when the number of interactions of the candidate media resource is less than a preset number.
If the candidate media resource belongs to the preset type of media resource, the adjusting module 302 may adjust the recommendation result of the candidate media resource. If the candidate media resource does not belong to the preset type of media resource, the adjusting module 302 may not adjust the recommendation result of the candidate media resource.
The adjustment module 302 may obtain a proportional parameter, an integral parameter, and a derivative parameter for adjusting the recommendation of the candidate media resource. The proportional parameter can represent the recommendation probability error closest to the current moment in the recommendation probability error sequence, the integral parameter can represent the accumulated error of the recommendation probability error sequence, and the differential parameter can represent the recommendation probability error change rate closest to the current moment in the recommendation probability error sequence.
The proportional parameter, the integral parameter, and the differential parameter may be determined based on a recommendation probability error sequence, where the recommendation probability error is determined based on recommendation probabilities respectively corresponding to a preset type of media resource and a non-preset type of media resource.
The adjustment module 302 may determine the control parameters for proportional, integral, and derivative regulation of the recommendation of the candidate media resource by: for each of at least one preset scheduling period prior to receiving the current request, performing the following: obtaining historical recommendation results of historical media resources in a plurality of historical requests in a preset scheduling period, wherein the historical media resources in the plurality of historical requests comprise a first media resource of a preset type and a second media resource of a non-preset type, and the historical recommendation results are obtained by prediction of a recommendation model; determining a recommendation probability of the recommendation model for the first media resource based on the historical recommendation result of the first media resource and determining a recommendation probability of the recommendation model for the second media resource based on the historical recommendation result of the second media resource; and determining a proportional parameter, an integral parameter and a differential parameter according to the recommendation probability of the first media resource and the second media resource of at least one preset scheduling period. .
The adjusting module 302 may rank each of the first media resources and each of the second media resources in the respective history requests according to the recommendation results of the first media resources and the second media resources, and calculate recommendation probabilities of the recommendation models for predicting recommendation results of the first media resources and the second media resources by comparing the ranking positions of the first media resources and the second media resources with a preset position. For example, the adjustment module 302 may rank each of the first media assets in the respective history request according to the historical recommendation results for the first media asset; calculating the recommendation probability of the recommendation model to the first media resource by comparing the sequencing position of the first media resource with a preset position; ranking each of the second media resources in the respective history request according to the history recommendation result of the second media resource; and calculating the recommendation probability of the recommendation model to the second media resource by comparing the sequencing position of the second media resource with a preset position.
Next, the adjusting module 302 may calculate a recommendation probability error of each preset scheduling period according to the recommendation probabilities of the first media resource and the second media resource of each preset scheduling period; accumulating the recommendation probability error of each preset scheduling period to obtain a recommendation probability error sequence; and respectively determining a proportional parameter, an integral parameter and a differential parameter by utilizing the recommended probability error sequence. For example, the adjusting module 302 may use the recommended probability error of the first preset scheduling period closest to the current request as a proportional parameter; taking the sum of each recommended probability error in the recommended probability error sequence as an integral parameter; and taking the difference between the recommended probability errors of the first preset scheduling period and the previous preset scheduling period of the first preset scheduling period as a differential parameter.
The recommending module 303 may select, according to the adjusted recommending result, a candidate media resource whose recommending result meets a preset condition from the plurality of candidate media resources as a media resource recommended to the user account, and send information of the selected candidate media resource to the user account.
The recommendation process has been described in detail above with reference to fig. 1 and 2, and will not be described in detail here.
Fig. 4 is a schematic structural diagram of a recommendation device of a hardware operating environment according to an embodiment of the present disclosure. Here, the recommendation device 400 may implement the functionality described above for efficiently recommending media assets.
As shown in fig. 4, the recommendation device 400 may include: a processing component 401, a communication bus 402, a network interface 403, an input output interface 404, a memory 405, and a power component 406. The communication bus 402 is used to implement, among other things, communication signals for connections between these components. The input/output interface 404 may include a video display (such as a liquid crystal display), a microphone and speaker, and a user interaction interface (such as a keyboard, mouse, touch input device, etc.), and optionally the input/output interface 404 may also include a standard wired interface, a wireless interface. The network interface 403 may optionally include a standard wired interface, a wireless interface (e.g., a wireless fidelity interface). Memory 405 may be a high speed random access memory or may be a stable non-volatile memory. The memory 405 may alternatively be a storage device separate from the aforementioned processing component 401.
Those skilled in the art will appreciate that the configuration shown in FIG. 4 is not limiting to the recommendation device 400 and may include more or fewer components than shown, or some components may be combined, or a different arrangement of components.
As shown in fig. 4, the memory 405, which is a storage medium, may include therein an operating system (such as a MAC operating system), a data storage module, a network communication module, a user interface module, a recommendation program, and a database.
In the recommendation device 400 shown in fig. 4, the network interface 403 is mainly used for data communication with an external device/terminal; the input/output interface 404 is mainly used for data interaction with a user account; the processing component 401 and the memory 405 in the recommendation device 400 may be provided in the recommendation device 400, and the recommendation device 400 may call the recommendation program stored in the memory 405 and various APIs provided by the operating system through the processing component 401, execute the recommendation method provided by the embodiments of the present disclosure, and the like.
The processing component 401 may include at least one processor, and the memory 405 has stored therein a set of computer-executable instructions that, when executed by the at least one processor, perform a recommendation method in accordance with an embodiment of the present disclosure. Further, processing component 401 may perform media resource recommendation processes and the like as described above. However, the above examples are merely exemplary, and the present disclosure is not limited thereto.
Further, the processing component 401 may receive a trained recommendation model from an external device and use the recommendation model to predict a recommendation for a media resource to be recommended, and then condition the recommendation for the media resource and determine whether to recommend the media resource to a user account based on the conditioned recommendation.
By way of example, the recommendation device 400 may be a PC computer, tablet device, personal digital assistant, smart phone, or other device capable of executing the set of instructions described above. Here, recommendation device 400 need not be a single electronic device, but can be any collection of devices or circuits capable of executing the above instructions (or sets of instructions) alone or in combination. The recommendation device 400 may also be part of an integrated control system or system manager, or may be configured as a portable electronic device that interfaces with local or remote (e.g., via wireless transmission).
In the recommendation device 400, the processing component 401 may include a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a programmable logic device, a special purpose processor system, a microcontroller, or a microprocessor. By way of example and not limitation, processing component 401 may also include an analog processor, a digital processor, a microprocessor, a multi-core processor, a processor array, a network processor, or the like.
The processing component 401 may execute instructions or code stored in a memory, wherein the memory 405 may also store data. Instructions and data may also be sent and received over a network via network interface 403, where network interface 403 may employ any known transmission protocol.
The memory 405 may be integral to the processor, e.g., having RAM or flash memory disposed within an integrated circuit microprocessor or the like. Further, memory 405 may comprise a stand-alone device, such as an external disk drive, storage array, or any other storage device that may be used by a database system. The memory and the processor may be operatively coupled or may communicate with each other, such as through an I/O port, a network connection, etc., so that the processor can read files stored in the memory.
According to an embodiment of the present disclosure, an electronic device may be provided. Fig. 5 is a block diagram of an electronic device 500 according to an embodiment of the disclosure, which may include at least one memory 502 and at least one processor 501, the at least one memory 502 storing a set of computer-executable instructions that, when executed by the at least one processor 501, perform a recommended method according to an embodiment of the disclosure.
The memory 502, which is a kind of storage medium, may include an operating system (e.g., a MAC operating system), a data storage module, a network communication module, a user interface module, a recommendation module, and a database.
The memory 502 may be integrated with the processor 501, for example, a RAM or flash memory may be disposed within an integrated circuit microprocessor or the like. Further, memory 502 may comprise a stand-alone device, such as an external disk drive, storage array, or any other storage device usable by a database system. The memory 502 and the processor 501 may be operatively coupled or may communicate with each other, such as through I/O ports, network connections, etc., so that the processor 501 can read files stored in the memory 502.
In addition, the electronic device 500 may also include a video display (such as a liquid crystal display) and a user interaction interface (such as a keyboard, mouse, touch input device, etc.). All components of the electronic device 500 may be connected to each other via a bus and/or a network.
By way of example, the electronic device 500 may be a PC computer, tablet device, personal digital assistant, smartphone, or other device capable of executing the set of instructions described above. Here, the electronic device 500 need not be a single electronic device, but can be any collection of devices or circuits that can execute the above instructions (or sets of instructions) individually or in combination. The electronic device 500 may also be part of an integrated control system or system manager, or may be configured as a portable electronic device that interfaces with local or remote (e.g., via wireless transmission).
Those skilled in the art will appreciate that the configuration shown in FIG. 5 is not intended to be limiting and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
According to an embodiment of the present disclosure, there may also be provided a computer-readable storage medium storing instructions that, when executed by at least one processor, cause the at least one processor to perform a recommendation method according to the present disclosure. Examples of computer-readable storage media herein include: read-only memory (ROM), random-access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random-access memory (RAM), dynamic random-access memory (DRAM), static random-access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD + R, CD-RW, CD + RW, DVD-ROM, DVD-R, DVD-RW, DVD + RW, DVD-RAM, BD-ROM, BD-R LTH, BD-RE, blu-ray or optical disk memory, hard Disk Drive (HDD), solid State Disk (SSD), card memory (such as a multimedia card, a Secure Digital (SD) card or an extreme digital (XD) card), magnetic tape, a floppy disk, a magneto-optical data storage device, an optical data storage device, a hard disk, a solid state disk, and any other device configured to store and provide computer programs and any associated data, data files and data structures in a non-transitory manner to a computer processor or computer such that the computer programs and any associated data processors are executed or computer programs. The computer program in the computer-readable storage medium described above can be run in an environment deployed in a computer apparatus, such as a client, a host, a proxy device, a server, and the like, and further, in one example, the computer program and any associated data, data files, and data structures are distributed across a networked computer system such that the computer program and any associated data, data files, and data structures are stored, accessed, and executed in a distributed fashion by one or more processors or computers.
According to an embodiment of the present disclosure, a computer program product may also be provided, in which instructions are executable by a processor of a computer device to perform the above-mentioned recommendation method.
Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice in the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
It will be understood that the present disclosure is not limited to the precise arrangements that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.
Claims (8)
1. A recommendation method, characterized in that the recommendation method comprises:
in response to receiving a current request of a user account for media resources, obtaining recommendation results of a plurality of candidate media resources by using a recommendation model;
for each of the candidate media resources, under the condition that the candidate media resource is determined to be a preset type of media resource, obtaining an adjustment parameter for adjusting a recommendation result of the candidate media resource and adjusting the recommendation result by using the adjustment parameter, wherein the interaction times of the preset type of media resource are less than preset times;
selecting candidate media resources meeting preset conditions from the candidate media resources according to the adjusted recommendation result as the media resources recommended to the user account,
wherein obtaining an adjustment parameter for adjusting the recommendation result of the candidate media resource comprises:
for each of at least one preset scheduling period prior to receiving the current request, performing the following: obtaining historical recommendation results of historical media resources in a plurality of historical requests in the preset scheduling period, wherein the historical media resources in the plurality of historical requests comprise first media resources of the preset type and second media resources of a non-preset type, and the historical recommendation results are obtained by prediction of the recommendation model; determining a recommendation probability of the recommendation model for the first media resource based on historical recommendations of the first media resource and determining a recommendation probability of the recommendation model for the second media resource based on historical recommendations of the second media resource;
and determining a proportional parameter, an integral parameter and a differential parameter in the adjusting parameters according to the recommendation probability of the first media resource and the second media resource of the at least one preset scheduling period.
2. The recommendation method according to claim 1, wherein the proportional parameter, the integral parameter and the differential parameter are determined based on a recommendation probability error sequence, and the recommendation probability error is determined based on recommendation probabilities respectively corresponding to the preset type of media resources and the non-preset type of media resources;
the proportional parameter represents a recommended probability error which is closest to the current moment in the recommended probability error sequence, the integral parameter represents an accumulated error of the recommended probability error sequence, and the differential parameter represents a recommended probability error change rate which is closest to the current moment in the recommended probability error sequence.
3. The recommendation method of claim 1, wherein determining the recommendation probability for the first media resource by the recommendation model based on the historical recommendation result for the first media resource and determining the recommendation probability for the second media resource by the recommendation model based on the historical recommendation result for the second media resource comprises:
according to the historical recommendation result of the first media resource, sequencing each of the first media resources in the respective historical request;
calculating the recommendation probability of the recommendation model to the first media resource by comparing the sequencing position of the first media resource with a preset position;
according to the historical recommendation result of the second media resource, sequencing each of the second media resources in the respective historical request;
and calculating the recommendation probability of the recommendation model to the second media resource by comparing the sequencing position of the second media resource with a preset position.
4. The recommendation method according to claim 1, wherein determining the proportional parameter, the integral parameter and the derivative parameter according to the recommendation probability of the first media resource and the second media resource of the at least one preset scheduling period comprises:
calculating a recommendation probability error of each preset scheduling period according to the recommendation probability of the first media resource and the second media resource of each preset scheduling period;
accumulating the recommended probability error of each preset scheduling period to obtain a recommended probability error sequence;
and respectively determining the proportional parameter, the integral parameter and the differential parameter by using the recommended probability error sequence.
5. The recommendation method according to claim 4, wherein determining the proportional parameter, the integral parameter, and the derivative parameter using the recommended probability error sequence, respectively, comprises:
taking a recommended probability error of a first preset scheduling period closest to the current request as the proportional parameter;
taking the sum of each recommended probability error in the recommended probability error sequence as the integral parameter;
and taking the difference between the recommended probability errors of the first preset scheduling period and the previous preset scheduling period of the first preset scheduling period as the differential parameter.
6. A recommendation device, characterized in that the recommendation device comprises:
a prediction module configured to obtain recommendation results for a plurality of candidate media resources using a recommendation model in response to receiving a current request for a media resource by a user account;
the adjusting module is configured to, for each of the plurality of candidate media resources, obtain an adjusting parameter for adjusting a recommendation result of the candidate media resource and adjust the recommendation result by using the adjusting parameter when the candidate media resource is determined to be a media resource of a preset type, where the number of interactions of the media resource of the preset type is less than a preset number of times;
a recommending module configured to select a candidate media resource meeting a preset condition from the plurality of candidate media resources as a media resource recommended to the user account according to the adjusted recommending result,
wherein the adjustment module is configured to, for each of at least one preset scheduling period prior to receiving the current request: obtaining historical recommendation results of historical media resources in a plurality of historical requests in the preset scheduling period, wherein the historical media resources in the plurality of historical requests comprise first media resources of the preset type and second media resources of a non-preset type, and the historical recommendation results are obtained by prediction of the recommendation model; determining a recommendation probability of the recommendation model for the first media resource based on historical recommendations of the first media resource and determining a recommendation probability of the recommendation model for the second media resource based on historical recommendations of the second media resource; and determining a proportional parameter, an integral parameter and a differential parameter in the adjusting parameters according to the recommendation probability of the first media resource and the second media resource of the at least one preset scheduling period.
7. An electronic device, comprising:
a processor;
a memory for storing the processor-executable instructions,
wherein the processor is configured to execute the instructions to implement the recommendation method of any one of claims 1-5.
8. A computer-readable storage medium, wherein instructions, when executed by a processor of an electronic device, enable the electronic device to perform the recommendation method of any of claims 1-5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211196512.0A CN115291503B (en) | 2022-09-29 | 2022-09-29 | Recommendation method, recommendation device, electronic device and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211196512.0A CN115291503B (en) | 2022-09-29 | 2022-09-29 | Recommendation method, recommendation device, electronic device and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115291503A CN115291503A (en) | 2022-11-04 |
CN115291503B true CN115291503B (en) | 2023-03-24 |
Family
ID=83833314
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211196512.0A Active CN115291503B (en) | 2022-09-29 | 2022-09-29 | Recommendation method, recommendation device, electronic device and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115291503B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116186417B (en) * | 2023-04-25 | 2023-07-04 | 北京阿帕科蓝科技有限公司 | Recommendation method, recommendation device, computer equipment and storage medium |
CN116719992A (en) * | 2023-05-26 | 2023-09-08 | 百度(中国)有限公司 | Resource recommendation method and device, electronic equipment and storage medium |
CN118013105A (en) * | 2023-08-01 | 2024-05-10 | 国家计算机网络与信息安全管理中心 | Push information generation method and device, electronic equipment and storage medium |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103631823A (en) * | 2012-08-28 | 2014-03-12 | 腾讯科技(深圳)有限公司 | Method and device for recommending media content |
CN112131411A (en) * | 2020-09-21 | 2020-12-25 | 腾讯科技(深圳)有限公司 | Multimedia resource recommendation method and device, electronic equipment and storage medium |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102724179B (en) * | 2012-05-25 | 2015-03-18 | 中国国际广播电台 | Information dissemination system based on integration of broadcasting network and Internet |
CN110020122B (en) * | 2017-10-16 | 2022-06-03 | Tcl科技集团股份有限公司 | Video recommendation method, system and computer readable storage medium |
CN110856003B (en) * | 2019-10-17 | 2021-11-02 | 网易(杭州)网络有限公司 | Live list pushing method and device, electronic equipment and storage medium |
CN114422826B (en) * | 2020-10-28 | 2023-06-27 | 腾讯科技(深圳)有限公司 | Multimedia content playing control method, device, equipment and medium |
CN114625893A (en) * | 2020-12-10 | 2022-06-14 | 北京达佳互联信息技术有限公司 | Media resource recall method, device, server and storage medium |
CN112948621A (en) * | 2021-01-30 | 2021-06-11 | 北京字节跳动网络技术有限公司 | Podcast recommendation method and device in music application |
-
2022
- 2022-09-29 CN CN202211196512.0A patent/CN115291503B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103631823A (en) * | 2012-08-28 | 2014-03-12 | 腾讯科技(深圳)有限公司 | Method and device for recommending media content |
CN112131411A (en) * | 2020-09-21 | 2020-12-25 | 腾讯科技(深圳)有限公司 | Multimedia resource recommendation method and device, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN115291503A (en) | 2022-11-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN115291503B (en) | Recommendation method, recommendation device, electronic device and storage medium | |
RU2720899C2 (en) | Method and system for determining user-specific content proportions for recommendation | |
CN110781321B (en) | Multimedia content recommendation method and device | |
US10868888B1 (en) | Method and apparatus for real-time personalization | |
CN110020094B (en) | Display method and related device for search results | |
US20170250930A1 (en) | Interactive content recommendation personalization assistant | |
RU2725659C2 (en) | Method and system for evaluating data on user-element interactions | |
RU2720952C2 (en) | Method and system for generating digital content recommendation | |
WO2017071251A1 (en) | Information pushing method and device | |
US9495645B2 (en) | Method and system of iteratively autotuning prediction parameters in a media content recommender | |
JP2018535497A (en) | Information recommendation method and apparatus | |
JP6737707B2 (en) | Method, apparatus and system for content recommendation | |
CN105431834A (en) | Creating personalized and continuous playlists for a content sharing platform based on user history | |
WO2015188349A1 (en) | Recommending of an item to a user | |
RU2714594C1 (en) | Method and system for determining parameter relevance for content items | |
CN111783810A (en) | Method and apparatus for determining attribute information of user | |
US20170154317A1 (en) | Enhanced content interest and consumption communities | |
Decker et al. | Sales forecasting in high‐technology markets: A utility‐based approach | |
JP7231322B2 (en) | LEARNING DEVICE, LEARNING METHOD, LEARNING PROGRAM AND PROGRAM | |
WO2018177303A1 (en) | Media content recommendation method, device, and storage medium | |
US20170365014A1 (en) | Systems, methods and non-transitory computer readable storage media for tracking and evaluating predictions regarding relationships | |
US11915256B2 (en) | Customized merchant price ratings | |
JP2024021038A (en) | Inter-domain shift recommendation system based on interest and preference of user | |
CN112269942B (en) | Method, device and system for recommending object and electronic equipment | |
US11631033B2 (en) | On-device training of personalized model |
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 |