CN113449177A - Information recommendation method and device, electronic equipment and storage medium - Google Patents

Information recommendation method and device, electronic equipment and storage medium Download PDF

Info

Publication number
CN113449177A
CN113449177A CN202010231254.XA CN202010231254A CN113449177A CN 113449177 A CN113449177 A CN 113449177A CN 202010231254 A CN202010231254 A CN 202010231254A CN 113449177 A CN113449177 A CN 113449177A
Authority
CN
China
Prior art keywords
user
target user
attention
candidate
rate
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.)
Pending
Application number
CN202010231254.XA
Other languages
Chinese (zh)
Inventor
宋琪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Dajia Internet Information Technology Co Ltd
Original Assignee
Beijing Dajia Internet Information Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Dajia Internet Information Technology Co Ltd filed Critical Beijing Dajia Internet Information Technology Co Ltd
Priority to CN202010231254.XA priority Critical patent/CN113449177A/en
Publication of CN113449177A publication Critical patent/CN113449177A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/01—Social networking

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Economics (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • Human Resources & Organizations (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The disclosure relates to an information recommendation method, an information recommendation device, electronic equipment and a storage medium, and relates to the technical field of internet. Acquiring specified behavior data generated by a target user in a user relationship development stage of the target user and a candidate user; according to the designated behavior data, recommending the target user by the user; the user relationship development stage comprises a stage when the user relationship between the target user and the candidate user is established and a stage after the user relationship is established; and the appointed behavior data is determined according to the historical behavior data prediction of the target user, and represents at least one of the attention behavior and the interaction behavior between the target user and the candidate user. The information recommendation method, the device, the electronic equipment and the storage medium disclosed by the invention are more beneficial to users settling on a platform to form an active community integrating content production and consumption and interaction among each other.

Description

Information recommendation method and device, electronic equipment and storage medium
Technical Field
The present disclosure relates to the field of internet technologies, and in particular, to an information recommendation method and apparatus, an electronic device, and a storage medium.
Background
In the social application product, a target user can be quickly and accurately found from a large number of platform users, and the social application product is a mission recommended by the user and is also one of main tools and means for establishing relationships among the platform users.
In the related art, user recommendations are mostly recommended based on factors in a relationship establishing stage so as to establish a mutual relationship between users, for example, recommendations are recommended based on user interests or familiar people or a relationship chain, and people who are interested in or familiar with the users can be found by adopting the recommendation mode.
For some platforms with decentralized content distribution as the main role, user recommendations are not only for establishing the correlation among users, but also for users to settle down to form an active community integrating content production and consumption and interaction among each other on the platform through the user recommendations. And the recommendation based on the factors in the relation establishment stage only enables the users to establish the mutual correlation, so that the users are difficult to ensure to precipitate and form an active community integrating content production and consumption and forming interaction among the users on the platform.
Disclosure of Invention
The present disclosure provides an information recommendation method, an information recommendation apparatus, an electronic device, and a storage medium, which at least solve a problem in the related art that it is difficult to ensure that a user settles on a platform because of user recommendation based on a factor at a relationship establishment stage.
The technical scheme of the disclosure is as follows:
according to a first aspect of the embodiments of the present disclosure, there is provided an information recommendation method, including:
acquiring specified behavior data generated by a target user in a user relationship development stage of the target user and a candidate user;
according to the specified behavior data, recommending the target user by the user;
the user relationship development stage comprises a stage when the user relationship between the target user and the candidate user is established and a stage after the user relationship is established;
and the specified behavior data is determined according to the historical behavior data prediction of the target user, and represents at least one of the attention behavior and the interaction behavior between the target user and the candidate user.
Optionally, before acquiring the specified behavior data generated by the target user in the user relationship development stage between the target user and the candidate user, the method further includes:
determining the candidate user associated with the target user based on the associated behavior data of the target user;
wherein the associated behavior data comprises at least one of historical behavior data, location data, and social relationship data of the target user.
Optionally, the acquiring specified behavior data generated by the target user in the user relationship development stage of the target user and the candidate user includes:
obtaining a behavior tag of at least one dimension between the target user and the candidate user according to at least one of data records of clicking, paying attention, deleting, paying attention, interacting and bidirectional attention generated by the target user in a user relationship development stage of the target user and the candidate user;
the behavior tag of the at least one dimension is at least one of click rate, attention rate, deletion rate, attention rejection rate, attention cancellation rate, interaction rate and bidirectional attention rate.
Optionally, the performing, to the target user, user recommendation includes:
obtaining a recommendation score between the target user and the candidate user according to the behavior label of at least one dimension between the target user and the candidate user;
and recommending the target user according to the recommendation score between the target user and the candidate user.
According to a second aspect of the embodiments of the present disclosure, there is provided an information recommendation apparatus including:
the acquisition unit is configured to execute acquisition of specified behavior data generated by a target user in a user relationship development stage of the target user and a candidate user;
the recommending unit is configured to execute user recommendation to the target user according to the specified behavior data;
the user relationship development stage comprises a stage when the user relationship between the target user and the candidate user is established and a stage after the user relationship is established;
and the specified behavior data is determined according to the historical behavior data prediction of the target user, and represents at least one of the attention behavior and the interaction behavior between the target user and the candidate user.
Optionally, the information recommendation device further includes:
a determining unit configured to perform determining the candidate user associated with the target user based on the associated behavior data of the target user;
wherein the associated behavior data comprises at least one of historical behavior data, location data, and social relationship data of the target user.
Optionally, the obtaining unit is configured to obtain a behavior tag of at least one dimension between the target user and the candidate user according to at least one of data records of click, attention, deletion, attention rejection, attention cancellation, interaction, and two-way attention generated by the target user in a user relationship development stage between the target user and the candidate user;
the behavior tag of the at least one dimension is at least one of click rate, attention rate, deletion rate, attention rejection rate, attention cancellation rate, interaction rate and bidirectional attention rate.
Optionally, the recommending unit is configured to perform at least one of data records of clicking, attention, deletion, attention rejection, attention cancellation, interaction and bidirectional attention of the target user on the candidate user to obtain a behavior tag of at least one dimension between the target user and the candidate user;
obtaining a recommendation score between the target user and the candidate user according to the behavior label of at least one dimension between the target user and the candidate user; and
and recommending the target user according to the recommendation score between the target user and the candidate user.
According to a third aspect of the embodiments of the present disclosure, there is provided an electronic apparatus including:
a processor;
a memory for storing the processor-executable instructions;
wherein the processor is configured to execute the instructions to implement the information recommendation method as described above.
According to a fourth aspect of embodiments of the present disclosure, there is provided a storage medium having instructions that, when executed by a processor of an electronic device, enable the electronic device to perform the information recommendation method as described above.
The technical scheme provided by the embodiment of the disclosure at least brings the following beneficial effects:
because the user recommendation is carried out on the target user according to the specified behavior data generated by the target user and the candidate user in the user relationship development stage, the user recommendation is not limited to the user relationship establishment, and therefore the method is more beneficial to the precipitation of the users on a platform to form an active community integrating content production and consumption and interaction among the users.
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 illustrating an information recommendation method according to an example embodiment.
FIG. 2 is a flow diagram illustrating user recommendation to a target user based on specified behavior data, according to an example embodiment.
Fig. 3 is a flow chart illustrating another information recommendation method according to an example embodiment.
FIG. 4 is a flow diagram illustrating training of a training model according to an exemplary embodiment.
FIG. 5 is a schematic diagram illustrating a first model and a second model according to an exemplary embodiment.
Fig. 6 is a block diagram illustrating an information recommendation apparatus according to an example embodiment.
Fig. 7 is a block diagram illustrating another information recommendation apparatus according to an example embodiment.
FIG. 8 is a block diagram illustrating an electronic device in accordance with an example embodiment.
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.
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 order to solve the problem that users are difficult to settle on a platform, the embodiment of the disclosure provides an information recommendation method, an information recommendation device, an electronic device and a storage medium, which can recommend users to target users according to specified behavior data generated by the target users and candidate users in a user relationship development stage, so that the method, the device, the electronic device and the storage medium are more beneficial to settling on the platform by the users to form an active community integrating content production and consumption and interaction among the users.
The information recommendation method provided by the embodiment of the present disclosure will be described in detail below.
The information recommendation method provided by the embodiments of the present disclosure may be applied to a server, and for convenience of description, the embodiments of the present disclosure all use the server as an execution subject for description except for specific description.
It is to be understood that the subject matter described herein is not to be construed as limiting the embodiments of the disclosure.
Specifically, the flow of the information recommendation method is shown in fig. 1, and may include the following steps:
in step S101, a candidate user associated with the target user is determined based on the associated behavior data of the target user.
In the embodiment of the disclosure, the target user may be any one of users in the social platform, and the candidate user is a user that may be recommended to the target user.
The associated behavior data of the target user may be, but is not limited to, historical behavior data, location data, social relationship data, and the like of the target user.
The historical behavior data may be historical clicks, messages, data records of interest, and the like of the target user, the location data may be Global Positioning System (GPS) coordinates when the target user currently logs in the social platform or location coordinates provided by a mobile network operator (such as mobile and telecommunication), and the social relationship data may refer to user information having a direct or indirect association relationship with the target user (such as an address book contact of the target user or another user having a common address book contact with the target user).
The candidate users are generally multiple and, in one embodiment of the present disclosure, may be determined from the list of objects of interest of the target user, the location data, and the social relationship data. For example, when determining the candidate user, a user having a part of the object list having common attention with the target user (i.e., a user who may have common preferences) may be selected as the candidate user. A portion of the list of target user clicked objects may be selected as candidate users. A portion of users in the same area (e.g., city) or within a certain distance from the target user may be selected as candidate users. A portion of users that are contact contacts with the target user or have common contact contacts with the target user may be selected as candidate users.
It is understood that the above is only an example, and in some other embodiments, the candidate user may have other determination manners, and the embodiments of the present disclosure are not limited in particular.
In step S103, specified behavior data generated by the target user in the user relationship development stage of the target user and the candidate user is obtained.
The user relationship development stage may include a stage during and after the user relationship between the target user and the candidate user is established, and the user relationship establishment may refer to clicking, paying attention, leaving messages and the like between the target user and the candidate user.
In the embodiment of the disclosure, the designated behavior data is determined according to the historical behavior data prediction of the target user, and represents at least one of the attention behavior and the interaction behavior between the target user and the candidate user.
Specifically, when acquiring and acquiring the specified behavior data generated by the target user in the user relationship development stage of the target user and the candidate user, the behavior tag of at least one dimension between the target user and the candidate user (that is, the specified behavior data generated by the target user in the user relationship development stage of the target user and the candidate user) can be obtained according to at least one of data records of clicking, concerning, deleting, rejecting concerning, cancelling concerning, interacting and bidirectional concerning generated by the target user in the user relationship development stage of the target user and the candidate user. The behavior tag of at least one dimension is at least one of click rate, attention rate, deletion rate, attention rejection rate, attention cancellation rate, interaction rate and bidirectional attention rate.
In the embodiment of the disclosure, a training model for calculating a behavior label of at least one dimension between a target user and the candidate user may be pre-established, and when calculating the behavior label of at least one dimension between the target user and the candidate user, at least one of data records of click, attention, deletion, attention rejection, attention cancellation, interaction and bidirectional attention generated by the target user in a user relationship development stage of the target user and the candidate user may be input to the training model, so as to output the behavior label of at least one dimension between the target user and the candidate user. The training process of the training model will be described in detail later, and will not be described herein again.
The behavior tag of at least one dimension is used for indicating the associated behavior of the target user and the candidate user in at least one dimension. In the embodiment of the disclosure, the historical behavior data includes data records of clicking, paying attention, deleting, rejecting attention, canceling attention, interacting, bidirectional attention and the like of the target user on the candidate user. Accordingly, the behavior tag of at least one dimension between the target user and the candidate user may include a click rate, an attention rate, a deletion rate, an attention rejection rate, an attention cancellation rate, an interaction rate, a two-way attention rate, and the like of the target user to the candidate user.
The click rate, the attention rate, the interaction rate and the two-way attention rate can reflect the possibility that the target user is interested in the candidate user to a certain degree, and the deletion rate, the attention rejection rate and the attention cancellation rate can reflect the possibility that the target user is not interested in the candidate user to a certain degree.
The click rate may be a probability that the target user clicks and browses the candidate user, and represents a possibility that the target user wants to know the candidate user. The attention rate may be a probability that the target user pays attention to the candidate user, indicating a likelihood that the target user is interested in and generates attention to the candidate user. The deletion rate may be the probability that the target user clicks the deletion candidate immediately, indicating the likelihood that the user dislikes the candidate user. The two-way interest rate may be a probability that two-way interest may be formed between the target user and the candidate user, indicating a likelihood of forming a more stable relationship between the target user and the candidate user. The attention cancellation rate may be a probability that the target user may generate attention cancellation after paying attention to the candidate user, and represents a possibility that the target user loses interest in the candidate user. The attention interaction rate may be a probability that the target user will interact after paying attention to the candidate user, and represents a possibility of forming close attention between users. The rate of declining attention may be a probability that recommending the same candidate user to the target user multiple times will not result in attention, indicating a likelihood that the user dislikes the candidate user.
A long-term stable relationship network is formed, which is beneficial to the stickiness of users and helps the development of a community platform. Therefore, in the embodiments of the present disclosure, the bidirectional attention rate, the cancellation attention rate, the attention interaction rate, and the rejection attention rate may be set as the behavior prediction within one period of time. That is, the behavior labels of at least one dimension between the target user and the candidate user include a click rate, an attention interaction rate within a predetermined time period (for convenience of description, hereinafter, referred to as an attention interaction rate), a bidirectional attention rate within a predetermined time period (for convenience of description, hereinafter, referred to as a bidirectional attention rate), a deletion rate, a cancellation attention rate within a predetermined time period (for convenience of description, hereinafter, referred to as a cancellation attention rate), and a rejection attention rate within a predetermined time period (for convenience of description, hereinafter, referred to as a rejection attention rate).
In the embodiment of the disclosure, the specified behavior data (behavior labels of at least one dimension between the user and the candidate users) characterizes the attention behavior and the interaction behavior of the target user between the candidate users. The characterization target user and the candidate user may include data such as click rate, attention rate, deletion rate, attention rejection rate, attention cancellation rate and the like of the target user on the candidate user, and the characterization target user and the candidate user may include data such as interaction rate, bidirectional attention rate and the like of the user and the candidate user.
Therefore, the specified behavior data generated in the user relationship development stage of the target user and the candidate user may include click rate, attention rate, deletion rate, attention rejection rate, attention cancellation rate, interaction rate and bidirectional attention rate, and the like.
For example, in the historical behavior data of the target user, the attention rate of the target user to the social class user is higher, and the deletion rate is lower, if the candidate user also belongs to the social class user, the click rate assignment in the attention behavior between the target user and the candidate user is higher, and the deletion rate assignment in the attention behavior between the target user and the candidate user is lower.
In step S105, user recommendation is made to the target user according to the specified behavior data.
As shown in fig. 2, according to the specified behavior data, the user recommendation to the target user includes the following steps:
in step S201, a recommendation score between the target user and the candidate user is obtained according to the behavior tag of at least one dimension between the target user and the candidate user.
The click rate, the attention rate, the interaction rate and the two-way attention rate can reflect the possibility that the target user is interested in the candidate user to a certain degree, and the deletion rate, the attention rejection rate and the attention cancellation rate can reflect the possibility that the target user is not interested in the candidate user to a certain degree.
That is to say, the larger the values corresponding to the behavior tags such as the click rate, the attention rate, the interaction rate, the two-way attention rate, and the like are, the higher the possibility that the target user is interested in the candidate user is, and the larger the values corresponding to the behavior tags such as the deletion rate, the attention rejection rate, the attention cancellation rate, and the like are, the higher the possibility that the target user is not interested in the candidate user is. Therefore, in the embodiment of the disclosure, the recommendation score between the target user and the candidate user can be directly obtained according to the values corresponding to the behavior tags, such as the click rate, the attention rate, the interaction rate, the two-way attention rate, the deletion rate, the attention rejection rate, the attention cancellation rate, and the like.
For example, the values corresponding to a plurality of behavior tags such as click rate, attention interaction rate, two-way attention rate and the like in the behavior tags of the at least one dimension may be added, and then the values corresponding to the behavior tags such as the corresponding deletion rate, attention cancellation rate, attention rejection rate and the like are subtracted, so that the obtained value is the recommendation score between the target user and each candidate user.
The focus of different users on the social platform is also different. Therefore, in the embodiment of the present disclosure, for different types of users, different strategies may be set when determining the recommended user of the target user, and different weighting coefficients are set for the click rate, the attention interaction rate, the two-way attention rate, the deletion rate, the attention cancellation rate, the attention rejection rate, and the like. The calculation formula of the recommendation Score between the target user and the candidate user may be expressed as Score ═ wftr*pftr+wctr*pctr+wbftr*pbftr+wiftr*piftr-wuftr*puftr-wdtr*pdtr-wrftr*prftrWherein Score represents the recommendation Score between the target user and a candidate user, wftrIndicates the rate of interest, pftrWeight coefficient, w, corresponding to the attention ratectrIndicates the click rate, pctrWeight coefficient, w, indicating the click ratebftrRepresenting a bidirectional interest rate, pbftrWeight coefficient, w, representing the correspondence of two-way interest ratesiftrIndicates the interaction rate of interest, piftrWeight coefficient, w, corresponding to the interaction rate of interestuftrIndicates the rate of canceling attention, puftrWeight coefficient, w, corresponding to the rate of interest cancellationdtrDenotes the deletion rate, pdtrWeight coefficient, w, indicating the erasure raterftrIndicates rejection of interest, prftrAnd a weight coefficient corresponding to the rejection attention rate.
For example, for a new user (e.g., a user whose registration time is less than 7 days), since the social relationship of the new user is simple, the user recommends that the triggering result is a hot start mode, so as to attract the attention of the user, in the setting of the weighting coefficients, the weighting coefficients corresponding to the attention rate and the click rate may be set relatively higher, and the other weighting coefficients may be set relatively lower. For a user with a preference to social contact (e.g., an old user with more interaction and mainly social contact), because the social contact user likes to like, comment, forward, and refresh, the main purpose of user recommendation is to help the user find a user who is more close and more likely to interact, so that the weight coefficient corresponding to the bidirectional attention rate and the weight coefficient corresponding to the click rate may be set relatively higher, and the other weight coefficients may be set relatively lower. For a user with a consumption preference (such as an old user with less interaction and mainly browsing), because the interaction is relatively less, the user generally only sees the works unsurpassed, commented and forwarded, the consumption quality requirement is relatively high, and the user recommendation mainly aims to help the user find a close and favorite user and simultaneously has a high negative feedback requirement, so that the weight coefficient corresponding to the attention rate cancellation and the weight coefficient corresponding to the deletion rate in the weight setting can be set to be higher, and other weight coefficients can be set to be relatively lower.
Different weight coefficients are respectively set for the behavior labels with different dimensions aiming at different types of users, so that interested users can be accurately recommended for the users with different types, the influence of negative recommendation on the users is reduced, and the viscosity of the platform is increased.
In step S203, user recommendation is performed to the target user according to the recommendation score between the target user and the candidate user.
When user recommendation is performed on a target user, whether recommendation scores between the target user and each candidate user exceed a preset threshold value or not can be judged, and if the recommendation scores exceed the preset threshold value, the candidate user corresponding to the recommendation scores is determined as the recommendation user of the target user and recommended to the target user. Or sorting the recommendation scores corresponding to all candidate users, selecting the candidate user corresponding to the part of recommendation scores at the top as the recommendation user of the target user, and recommending the candidate user to the target user.
Please refer to fig. 3, which is a flowchart of another information recommendation method according to an embodiment of the disclosure, including the following steps:
in step S301, based on the historical designated behavior data generated by the historical target user in the relationship development stage between the historical target user and the historical recommended user as sample input, the historical behavior label of at least one dimension generated after the historical recommended user is recommended to the historical target user as sample output is trained, and a training model is obtained.
In the embodiment of the disclosure, before the specified behavior data generated by the target user in the user relationship development stage of the target user and the candidate user is input to the training model to output the behavior label of at least one dimension between the target user and the candidate user, the method further includes a process of training the information recommendation model. In the subsequent process of recommending recommended users to target users on the social platform, the information recommendation model does not need to be trained every time, or the training model can be periodically updated on the basis of newly acquired training samples on the social platform, so that the prediction accuracy of the training model is improved, and the accuracy in the process of recommending social users is further improved.
As shown in fig. 4, the training process of the training model may include the following steps:
in step S401, history designated behavior data generated by the history target user in the relationship development stage between the history target user and the history recommending user is used as a sample to be input into the first model, and a first result is obtained.
The history target user refers to a user to which a recommending user is added, and the history recommending user refers to a user who is recommended to the history target user and added by the history target user. The historical designated behavior data generated by the historical target user in the relation development stage of the historical target user and the historical recommendation user may be data records of clicking, concerning, deleting, rejecting to be concerned, cancelling to be concerned, interacting, paying attention in two directions and the like of the historical recommendation user recommended by the historical target user, and the embodiment of the present disclosure is not particularly limited.
The first result represents multi-dimensional behavior tags generated after history recommending users are recommended to history target users under the condition that interaction between various attention and interaction behaviors (including clicking, attention, deleting, rejecting attention, canceling attention, interacting, two-way attention and the like) is not considered. The multidimensional behavior tags may include, but are not limited to, a click rate, an attention rate, an interaction rate, a two-way interaction rate, and the like for indicating that a positive social relationship behavior is generated, and a deletion rate, an attention cancellation rate, an attention rejection rate, and the like for indicating that a negative social relationship behavior is generated. If the history target user has a certain attention or interaction behavior to the history recommending user, the value of the behavior tag corresponding to the dimension is 1, and if the history target user has no certain attention or interaction behavior to the history recommending user, the value of the behavior tag corresponding to the dimension is 0. For example, if a history target user looks at a history recommended user but does not pay attention to the history recommended user, the click rate is 1 and the attention rate is 0 in the corresponding multi-dimensional behavior tag.
In step S403, the first result is used as an input of the second model, and a historical behavior tag of at least one dimension generated after the historical recommended user is recommended to the historical target user is used as a sample output for training, so as to obtain a training model.
In the embodiment of the present disclosure, the first model may adopt a Deep Neural Network (DNN) and the second model may adopt a Bayesian Network (BN). As shown in fig. 5, the double-tower deep neural network includes an embedding layer, a sharing layer, and an output layer, and since the bidirectional interest rate, the interest cancellation rate, the interest interaction rate, and the interest rejection rate are set as behavior predictions within a time period, and the click rate, the interest rate, and the deletion rate are behavior predictions at time points, as shown in fig. 5, the behavior predictions within the time period and at the time points may respectively correspond to different embedding layers, sharing layers, and output layers.
During training, a training sample is firstly input into an embedding layer, and an embedding vector embedding is obtained by the embedding layer based on historical specified behavior data generated by a historical target user in the training sample in a relation development stage of the historical target user and a historical recommendation user, wherein the embedding vector embedding is a vector constructed according to a mapping relation, and the mapping relation can be used for mapping information in a space into a vector space so as to construct respective network expressions. And then, the embedded vectors output by the embedded layer are superposed and input into the shared layer, the input embedded vectors are fused by the shared layer and then connected to the output layer, and the probability prediction result of at least one dimension of historical behavior labels generated by the historical target user on the historical recommended user is output after the processing of the output layer, wherein the probability prediction result is the result without considering the mutual influence between the attention and the interaction behaviors.
In practical situations, the behavior of some dimensions may depend on the behavior of other dimensions, for example, focus may depend on click behavior, two-way focus behavior may depend on focus behavior, and cancel focus and interaction behavior may depend on focus and two-way focus behavior. Therefore, after the output layer outputs the probability prediction result of the historical behavior label of at least one dimension without considering the mutual influence between the attention and the interactive behaviors, for the attention or the interactive behaviors which are influenced by the behaviors corresponding to the historical behavior labels of other dimensions in the Bayesian network, the probability prediction result corresponding to the attention or the interactive behaviors is fused with the probability prediction result of the historical behavior labels of other dimensions which are depended on the attention or the interactive behaviors, and finally, the probability prediction result of the historical behavior label of at least one dimension, which is generated by the historical target user to the historical recommendation user under the condition of considering the mutual influence between the attention and the interactive behaviors, is output. And finally, with a minimum loss function as a target, continuously adjusting parameters of the double-tower neural network based on the labels corresponding to the training samples and the prediction result output by the model, and finally training to obtain a training model for predicting the probability of the attention or the interaction behavior of the user to at least one dimension of the user of the recommended object.
In step S303, a candidate user associated with the target user is determined based on the associated behavior data of the target user.
In step S305, the designated behavior data generated by the target user in the user relationship development stage of the target user and the candidate user is obtained.
In step S307, user recommendation is made to the target user according to the specified behavior data.
In the embodiment of the disclosure, the specified behavior data generated by the target user in the user relationship development stage of the target user and the candidate user can be input into a pre-trained training model to output the behavior label of at least one dimension between the target user and the candidate user, then the recommendation score between the target user and the candidate user is obtained according to the behavior label of at least one dimension between the target user and the candidate user, and the user recommendation is performed to the target user according to the recommendation score. The appointed behavior data can represent clicking, paying attention, deleting attention, rejecting attention, paying attention, interacting, two-way attention and the like between the target user and the candidate user, so that the problems of the system in all links of user recommendation are considered, and compared with the problem that the conventional user recommendation value only stays in the relationship establishment stage, the scheme disclosed by the invention not only considers the behaviors of the user in the recommendation process of clicking, paying attention or deleting the recommended user, but also considers whether the user can form two-way attention or generate interactive behaviors within a period of time or not, and repeatedly recommends the behaviors which are not paid attention or cancel the attention after the attention is paid, so that a more stable relationship can be formed between the recommended user and the recommended user, the method is more beneficial to the users in platform precipitation to form integrated content production and consumption and form interactive active communities, and guides the users to form better quality, more interactive active communities, The method has the advantages of more stable and active two-way interaction friendly relation and reduction of the influence of negative recommendation on the user. Secondly, according to the scheme, when the training model is established, behavior labels of all dimensions are not treated under the same conditions, but are modeled through a Bayesian network according to the dependency relationship among the behavior labels of all dimensions in the using process of a user, and the recommendation accuracy is obviously superior to that of the traditional estimation model. In addition, different weight coefficients are respectively set for the tags used for indicating that the positive social relation behaviors are generated and the tags used for indicating that the negative social relation behaviors are generated aiming at different types of users, so that users interested in the tags can be accurately recommended for the users of different types, the influence of the negative recommendation on the users is further reduced, and the viscosity of the platform is increased.
Please refer to fig. 6, which is a block diagram of an information recommendation apparatus 600 according to an embodiment of the disclosure. As shown in fig. 6, the information recommendation apparatus 600 includes a determination unit 601, an acquisition unit 603, and a recommendation unit 605.
The determining unit 601 is configured to perform determining the candidate user associated with the target user based on the associated behavior data of the target user.
Wherein the associated behavior data comprises at least one of historical behavior data, location data, and social relationship data of the target user.
The obtaining unit 603 is configured to perform obtaining of specified behavior data generated by the target user in a user relationship development stage of the target user and the candidate user.
The user relationship development stage comprises stages of establishing the user relationship between the target user and the candidate user and after establishing the user relationship.
And the specified behavior data is determined according to the historical behavior data prediction of the target user, and represents at least one of the attention behavior and the interaction behavior between the target user and the candidate user.
Specifically, the fetching unit 603 is configured to obtain a behavior tag of at least one dimension between the target user and the candidate user according to at least one of data records of click, attention, deletion, attention rejection, attention cancellation, interaction, and two-way attention generated by the target user in a user relationship development stage of the target user and the candidate user. The behavior tag of the at least one dimension is at least one of click rate, attention rate, deletion rate, attention rejection rate, attention cancellation rate, interaction rate and bidirectional attention rate.
The recommending unit 605 is configured to perform user recommendation to the target user according to the specified behavior data.
Specifically, the recommending unit 605 is configured to execute behavior tagging according to at least one dimension between the target user and the candidate user to obtain a recommendation score between the target user and the candidate user; and recommending users to the target users according to the recommendation scores between the target users and the candidate users.
Further, the recommending unit 605 may be further configured to perform obtaining the recommendation score between the target user and the candidate user according to the behavior tag of at least one dimension between the target user and the candidate user and the weighting coefficient corresponding to the behavior tag of the at least one dimension.
Referring to fig. 7, in an embodiment of the present disclosure, the information recommendation apparatus 600 further includes a training unit 607.
The training unit 607 is configured to perform training based on the historical designated behavior data generated by the historical target user in the relationship development stage between the historical target user and the historical recommendation user as sample input and the historical behavior label of at least one dimension generated after the historical recommendation user is recommended to the historical target user as sample output, so as to obtain a training model.
Specifically, the training unit 607 is configured to perform inputting the historical assigned behavior data generated by the historical target user in the relationship development stage between the historical target user and the historical recommendation user into the first model as a sample, and obtain a first result. And taking the first result as the input of the second model, recommending at least one dimension historical behavior label generated after the historical recommendation user is recommended to the historical target user as sample output, and training to obtain a training model.
With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.
FIG. 8 is a block diagram illustrating an electronic device in accordance with an example embodiment. Referring to fig. 8, the electronic device includes a processing component 801 that further includes one or more processors, and memory resources, represented by memory 802, for storing instructions, such as application programs, that are executable by the processing component 802. The application programs stored in memory 802 may include one or more modules that each correspond to a set of instructions. Further, the processing component 801 is configured to execute instructions to perform the recommendation method described above. For example, the recommendation method includes the steps of:
acquiring specified behavior data generated by a target user in a user relationship development stage of the target user and a candidate user;
according to the specified behavior data, recommending the target user by the user;
the user relationship development stage comprises a stage when the user relationship between the target user and the candidate user is established and a stage after the user relationship is established;
and the specified behavior data is determined according to the historical behavior data prediction of the target user, and represents at least one of the attention behavior and the interaction behavior between the target user and the candidate user.
The electronic device may also include a power component 803 configured to perform power management of the electronic device, a wired or wireless network interface 804 configured to connect the server to a network, and an input/output (I/O) interface 805. The electronic device may operate based on an operating system stored in memory 802, such as Windows Server, Mac OSXTM, UnixTM, LinuxTM, FreeBSDTM, or the like.
Through the electronic equipment disclosed by the embodiment of the disclosure, the specified behavior data generated by the target user in the user relationship development stage of the target user and the candidate user can be input into a pre-trained training model to output the behavior label of at least one dimension between the target user and the candidate user, then the recommendation score between the target user and the candidate user is obtained according to the behavior label of at least one dimension between the target user and the candidate user, and the user recommendation is carried out on the target user according to the recommendation score. The appointed behavior data can represent clicking, paying attention, deleting attention, rejecting attention, paying attention, interacting, two-way attention and the like between the target user and the candidate user, so that the problems of the system in all links of user recommendation are considered, and compared with the problem that the conventional user recommendation value only stays in the relationship establishment stage, the scheme disclosed by the invention not only considers the behaviors of the user in the recommendation process of clicking, paying attention or deleting the recommended user, but also considers whether the user can form two-way attention or generate interactive behaviors within a period of time or not, and repeatedly recommends the behaviors which are not paid attention or cancel the attention after the attention is paid, so that a more stable relationship can be formed between the recommended user and the recommended user, the method is more beneficial to the users in platform precipitation to form integrated content production and consumption and form interactive active communities, and guides the users to form better quality, more interactive active communities, The method has the advantages of more stable and active two-way interaction friendly relation and reduction of the influence of negative recommendation on the user. Secondly, according to the scheme, when the training model is established, behavior labels of all dimensions are not treated under the same conditions, but are modeled through a Bayesian network according to the dependency relationship among the behavior labels of all dimensions in the using process of a user, and the recommendation accuracy is obviously superior to that of the traditional estimation model. In addition, different weight coefficients are respectively set for the tags used for indicating that the positive social relation behaviors are generated and the tags used for indicating that the negative social relation behaviors are generated aiming at different types of users, so that users interested in the tags can be accurately recommended for the users of different types, the influence of the negative recommendation on the users is further reduced, and the viscosity of the platform is increased.
In an exemplary embodiment, a storage medium comprising instructions, such as memory 802 comprising instructions, executable by a processor of apparatus 600 to perform the above-described recommendation method is also provided. For example, the recommendation method comprises the steps of:
acquiring specified behavior data generated by a target user in a user relationship development stage of the target user and a candidate user;
according to the specified behavior data, recommending the target user by the user;
the user relationship development stage comprises a stage when the user relationship between the target user and the candidate user is established and a stage after the user relationship is established;
and the specified behavior data is determined according to the historical behavior data prediction of the target user, and represents at least one of the attention behavior and the interaction behavior between the target user and the candidate user.
Alternatively, the storage medium may be a non-transitory computer readable storage medium, which may be, for example, a ROM, a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
Through the storage medium disclosed by the embodiment of the disclosure, the specified behavior data generated by the target user in the user relationship development stage of the target user and the candidate user can be input into a pre-trained training model to output the behavior label of at least one dimension between the target user and the candidate user, then the recommendation score between the target user and the candidate user is obtained according to the behavior label of at least one dimension between the target user and the candidate user, and the user recommendation is performed on the target user according to the recommendation score. The appointed behavior data can represent clicking, paying attention, deleting attention, rejecting attention, paying attention, interacting, two-way attention and the like between the target user and the candidate user, so that the problems of the system in all links of user recommendation are considered, and compared with the problem that the conventional user recommendation value only stays in the relationship establishment stage, the scheme disclosed by the invention not only considers the behaviors of the user in the recommendation process of clicking, paying attention or deleting the recommended user, but also considers whether the user can form two-way attention or generate interactive behaviors within a period of time or not, and repeatedly recommends the behaviors which are not paid attention or cancel the attention after the attention is paid, so that a more stable relationship can be formed between the recommended user and the recommended user, the method is more beneficial to the users in platform precipitation to form integrated content production and consumption and form interactive active communities, and guides the users to form better quality, more interactive active communities, The method has the advantages of more stable and active two-way interaction friendly relation and reduction of the influence of negative recommendation on the user. Secondly, according to the scheme, when the training model is established, behavior labels of all dimensions are not treated under the same conditions, but are modeled through a Bayesian network according to the dependency relationship among the behavior labels of all dimensions in the using process of a user, and the recommendation accuracy is obviously superior to that of the traditional estimation model. In addition, different weight coefficients are respectively set for the tags used for indicating that the positive social relation behaviors are generated and the tags used for indicating that the negative social relation behaviors are generated aiming at different types of users, so that users interested in the tags can be accurately recommended for the users of different types, the influence of the negative recommendation on the users is further reduced, and the viscosity of the platform is increased.
The embodiment of the disclosure also provides a computer program product, which includes instructions, when the computer program product is executed by a computer, the instructions make the computer execute the recommendation method. For example, the recommendation method comprises the steps of:
acquiring specified behavior data generated by a target user in a user relationship development stage of the target user and a candidate user;
according to the specified behavior data, recommending the target user by the user;
the user relationship development stage comprises a stage when the user relationship between the target user and the candidate user is established and a stage after the user relationship is established;
and the specified behavior data is determined according to the historical behavior data prediction of the target user, and represents at least one of the attention behavior and the interaction behavior between the target user and the candidate user.
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 within 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 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 (10)

1. An information recommendation method, comprising:
acquiring specified behavior data generated by a target user in a user relationship development stage of the target user and a candidate user;
according to the specified behavior data, recommending the target user by the user;
the user relationship development stage comprises a stage when the user relationship between the target user and the candidate user is established and a stage after the user relationship is established;
and the specified behavior data is determined according to the historical behavior data prediction of the target user, and represents at least one of the attention behavior and the interaction behavior between the target user and the candidate user.
2. The method of claim 1, wherein prior to obtaining the specified behavior data generated by the target user during the user relationship development stage of the target user and the candidate user, the method further comprises:
determining the candidate user associated with the target user based on the associated behavior data of the target user;
wherein the associated behavior data comprises at least one of historical behavior data, location data, and social relationship data of the target user.
3. The method according to claim 1, wherein the obtaining of the specified behavior data generated by the target user in the user relationship development stage of the target user and the candidate user comprises:
obtaining a behavior tag of at least one dimension between the target user and the candidate user according to at least one of data records of clicking, paying attention, deleting, paying attention, interacting and bidirectional attention generated by the target user in a user relationship development stage of the target user and the candidate user;
the behavior tag of the at least one dimension is at least one of click rate, attention rate, deletion rate, attention rejection rate, attention cancellation rate, interaction rate and bidirectional attention rate.
4. The method of claim 3, wherein making user recommendations to the target user based on the specified behavior data comprises:
obtaining a recommendation score between the target user and the candidate user according to the behavior label of at least one dimension between the target user and the candidate user;
and recommending the target user according to the recommendation score between the target user and the candidate user.
5. An information recommendation apparatus, comprising:
the acquisition unit is configured to execute acquisition of specified behavior data generated by a target user in a user relationship development stage of the target user and a candidate user;
the recommending unit is configured to execute user recommendation to the target user according to the specified behavior data;
the user relationship development stage comprises a stage when the user relationship between the target user and the candidate user is established and a stage after the user relationship is established;
and the specified behavior data is determined according to the historical behavior data prediction of the target user, and represents at least one of the attention behavior and the interaction behavior between the target user and the candidate user.
6. The apparatus of claim 5, further comprising:
a determining unit configured to perform determining the candidate user associated with the target user based on the associated behavior data of the target user;
wherein the associated behavior data comprises at least one of historical behavior data, location data, and social relationship data of the target user.
7. The apparatus according to claim 5, wherein the obtaining unit is configured to obtain a behavior tag of at least one dimension between the target user and the candidate user according to at least one of data records of click, attention, deletion, attention rejection, attention cancellation, interaction and two-way attention generated by the target user in a user relationship development stage of the target user and the candidate user;
the behavior tag of the at least one dimension is at least one of click rate, attention rate, deletion rate, attention rejection rate, attention cancellation rate, interaction rate and bidirectional attention rate.
8. The apparatus according to claim 7, wherein the recommending unit is configured to perform obtaining a behavior tag of at least one dimension between the target user and the candidate user according to at least one of data records of clicking, attention focusing, deleting, attention rejecting, attention canceling, interaction and bidirectional attention of the target user on the candidate user;
obtaining a recommendation score between the target user and the candidate user according to the behavior label of at least one dimension between the target user and the candidate user; and
and recommending the target user according to the recommendation score between the target user and the candidate user.
9. 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 information recommendation method of any one of claims 1 to 4.
10. A storage medium in which instructions, when executed by a processor of an electronic device, enable the electronic device to perform the information recommendation method of any one of claims 1 to 4.
CN202010231254.XA 2020-03-27 2020-03-27 Information recommendation method and device, electronic equipment and storage medium Pending CN113449177A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010231254.XA CN113449177A (en) 2020-03-27 2020-03-27 Information recommendation method and device, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010231254.XA CN113449177A (en) 2020-03-27 2020-03-27 Information recommendation method and device, electronic equipment and storage medium

Publications (1)

Publication Number Publication Date
CN113449177A true CN113449177A (en) 2021-09-28

Family

ID=77807980

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010231254.XA Pending CN113449177A (en) 2020-03-27 2020-03-27 Information recommendation method and device, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN113449177A (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104281622A (en) * 2013-07-11 2015-01-14 华为技术有限公司 Information recommending method and information recommending device in social media
CN107835120A (en) * 2017-11-03 2018-03-23 上海掌门科技有限公司 For providing the method and apparatus to release news of target good friend
CN107909428A (en) * 2017-11-01 2018-04-13 平安科技（深圳）有限公司 Electronic device, Products Show method and computer-readable recording medium
CN110209952A (en) * 2018-12-18 2019-09-06 腾讯科技（深圳）有限公司 Information recommendation method, device, equipment and storage medium
CN110225122A (en) * 2019-06-13 2019-09-10 广州酷狗计算机科技有限公司 A kind of information push method, device, equipment and storage medium
CN110263242A (en) * 2019-01-04 2019-09-20 腾讯科技（深圳）有限公司 Content recommendation method, device, computer readable storage medium and computer equipment

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104281622A (en) * 2013-07-11 2015-01-14 华为技术有限公司 Information recommending method and information recommending device in social media
CN107909428A (en) * 2017-11-01 2018-04-13 平安科技（深圳）有限公司 Electronic device, Products Show method and computer-readable recording medium
CN107835120A (en) * 2017-11-03 2018-03-23 上海掌门科技有限公司 For providing the method and apparatus to release news of target good friend
CN110209952A (en) * 2018-12-18 2019-09-06 腾讯科技（深圳）有限公司 Information recommendation method, device, equipment and storage medium
CN110263242A (en) * 2019-01-04 2019-09-20 腾讯科技（深圳）有限公司 Content recommendation method, device, computer readable storage medium and computer equipment
CN110225122A (en) * 2019-06-13 2019-09-10 广州酷狗计算机科技有限公司 A kind of information push method, device, equipment and storage medium

Similar Documents

Publication Publication Date Title
CN110458663B (en) Vehicle recommendation method, device, equipment and storage medium
CN110070391B (en) Data processing method and device, computer readable medium and electronic equipment
US11093992B2 (en) Smart matching for real estate transactions
Bouneffouf et al. A contextual-bandit algorithm for mobile context-aware recommender system
CN107463701B (en) Method and device for pushing information stream based on artificial intelligence
US11113745B1 (en) Neural contextual bandit based computational recommendation method and apparatus
CN107330715B (en) Method and device for selecting picture advertisement material
CN109816483B (en) Information recommendation method and device and readable storage medium
CN108446382B (en) Method and apparatus for pushed information
JP6911603B2 (en) How to generate predictive models for the categories of facilities visited by users, programs, server equipment, and processing equipment
WO2017116931A2 (en) Task similarity clusters for worker assessment
US10438114B1 (en) Recommending content using neural networks
CN110275746B (en) Page display control method and device
US7895149B2 (en) System for opinion reconciliation
CN110971659A (en) Recommendation message pushing method and device and storage medium
CN107463580B (en) Click rate estimation model training method and device and click rate estimation method and device
US20160132892A1 (en) Method and system for estimating customer satisfaction
US20210035183A1 (en) Method and system for a recommendation engine utilizing progressive labeling and user content enrichment
KR20130035660A (en) Recommendation system and method
US20190303771A1 (en) Inferred profiles on online social networking systems using network graphs
CN110046642B (en) Computer-implemented user intention prediction method, device and equipment
CN107408125A (en) For inquiring about the image of answer
US20240144328A1 (en) Automatic rule generation for next-action recommendation engine
CN113869931A (en) Advertisement putting strategy determining method and device, computer equipment and storage medium
KR20110096488A (en) Collaborative networking with optimized inter-domain information quality assessment

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