CN110046258B - Method and device for determining user association metric values for entities - Google Patents
Method and device for determining user association metric values for entities Download PDFInfo
- Publication number
- CN110046258B CN110046258B CN201910176407.2A CN201910176407A CN110046258B CN 110046258 B CN110046258 B CN 110046258B CN 201910176407 A CN201910176407 A CN 201910176407A CN 110046258 B CN110046258 B CN 110046258B
- Authority
- CN
- China
- Prior art keywords
- user
- entity
- value
- feature vector
- feature
- 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
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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- 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
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Finance (AREA)
- General Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Physics & Mathematics (AREA)
- Animal Behavior & Ethology (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present disclosure provides methods and apparatus for determining a user association metric value for an entity. The method comprises the following steps: updating, for each entity present at a user, an entity feature vector for the entity based on a first entity feature variation value for the entity, the first entity feature variation value received from a neighbor user of the user, the neighbor user determined from a predetermined collaborative user group including the user; and determining a user association metric value for each entity based on the user preference feature vector of the user and the updated entity feature vector of each entity.
Description
Technical Field
The present disclosure relates generally to the field of computer technology, and more particularly, to a method and apparatus for determining a user-associated metric value for an entity.
Background
With the development of modern technology, businesses in various fields are becoming more and more intelligent. When people perform various activities (such as shopping, browsing information, listening to music and the like) through the Internet, instead of simply searching for target entities which are favored by the users, platforms (such as application software, website systems and the like) for performing various activities of people recommend target entities which are possibly favored by the users to the users according to big data. In order to recommend the target entity to the user, the association degree between the user and each entity needs to be predicted, and recommendation is further performed based on the association degree. Systems that recommend entities to a user based on this degree of association are commonly referred to as recommendation systems.
Prior art recommendation systems typically collect a chinese (centered) training approach to determine the degree of association between a user and an entity (e.g., a user's score for an entity). And the recommending system can recommend the entity with the highest score to the user. Fig. 1 is a schematic diagram of a recommendation system 100 in the prior art. As shown in fig. 1, in the prior art, a server 130 of the recommendation system 100 is generally first constructed, operation behavior (such as purchasing, clicking, scoring, etc.) data 120 of each user 110 on each entity is obtained by the server 130, and then a scoring model 150 is trained by using summarized data 140 of the behavior data, and the scoring model 150 is further used to construct the recommendation system 100.
In the recommendation system in the prior art, data are stored in a server in a centralized manner, and a scoring model is trained in a centralized manner in the server.
Disclosure of Invention
In view of the foregoing, the present disclosure provides a method and apparatus for determining a user association metric value for an entity. By using the method and the device, the user association metric values of the entities corresponding to the users can be predicted in cooperation with the users, so that uploading of user data to a server is not needed, and the user association metric values are determined at the users, thereby not only avoiding leakage of user privacy, but also saving storage cost and operation cost.
According to one aspect of the present disclosure, there is provided a method for determining a user association metric value for an entity, comprising: updating, for each entity present at a user, an entity feature vector for the entity based on a first entity feature variation value for the entity, the first entity feature variation value received from a neighbor user of the user, the neighbor user determined from a predetermined collaborative user group including the user; and determining a user association metric value for each entity based on the user preference feature vector of the user and the updated entity feature vector of each entity.
Optionally, in one example, the method may further include: when the operation behaviors of the user for the entities are acquired, determining a user preference characteristic change value of the user and a second entity characteristic change value of the entity; and updating the user preference feature vector for the user based on the user preference feature variation value. Updating the entity feature vector for the entity based on the first entity feature variation value for the entity may include: the entity feature vector of the entity is updated based on the second entity feature variation value of the entity and/or the first entity feature variation value of the entity. Determining the user association metric value for each entity based on the user preference feature vector of the user and the updated entity feature vector of each entity may include: and determining a user association metric value for each entity based on the updated user preference feature vector and the updated entity feature vector of each entity.
Optionally, in one example, the method may further include: and transmitting the second entity characteristic change value and/or the first entity characteristic change value to at least one neighbor user of the user, wherein the first entity characteristic change value is provided with a time stamp, and the time stamp indicates the sending time of the first entity characteristic change value.
Optionally, in one example, the sending of the first entity characteristic change value to at least one neighbor user of the user may be based on a continue delivery indication.
Optionally, in one example, the continued delivery indication may be determined based on a delivery times constraint of the first entity characteristic change value.
Optionally, in one example, determining the user preference feature variation value of the user and the second entity feature variation value of the respective entity may include: based on the loss function, a user preference feature variation value of the user and a second entity feature variation value of the respective entity are determined. Wherein the penalty function is a function of a user association metric value for an entity corresponding to the user, the user preference feature vector, and the entity feature vector, the user association metric value for an entity corresponding to the user being determined based on a user's operational behavior for the entity.
Optionally, in one example, the collaborative user group is determined based on a distance between individual users.
Alternatively, in one example, the user preference feature vector and the entity feature vector may be hash value vectors.
Alternatively, in one example, the hash value vector may be a binary vector.
Optionally, in one example, the entity may include at least one of: an article; a user; and information.
Optionally, in one example, the method may further include: recommending the target entity to the user based on the determined user association metric values for the entities.
According to another aspect of the present disclosure, there is provided an apparatus for determining a user association metric value for an entity, comprising: an updating unit configured to update, for each entity present at a user, an entity feature vector of the entity based on a first entity feature variation value of the entity, the first entity feature variation value being received from a neighbor user of the user, the neighbor user being determined from a predetermined collaborative user group including the user; and a user association metric value determination unit configured to determine a user association metric value for each entity based on the user preference feature vector of the user and the updated entity feature vector of each entity.
Optionally, in one example, the apparatus may further include: and a feature change value determining unit configured to determine a user preference feature change value of the user and a second entity feature change value of the entity when the operation behavior of the user with respect to the respective entities is acquired. The updating unit may be further configured to: updating the user preference feature vector of the user based on the user preference feature change value; and updating the entity feature vector of the entity based on the second entity feature variation value of the entity and/or the first entity feature variation value of the entity. The user association metric value determination unit is configured to: and determining a user association metric value for each entity based on the updated user preference feature vector and the updated entity feature vector of each entity.
Optionally, in one example, the apparatus may further include: and the feature change value sending unit is configured to send the second entity feature change value and/or the first entity feature change value to at least one neighbor user of the user.
Alternatively, in one example, the feature change value determining unit may be configured to: based on the loss function, a user preference feature variation value of the user and a second entity feature variation value of the respective entity are determined. Wherein the penalty function is a function of a user association metric value for an entity corresponding to the user, the user preference feature vector, and the entity feature vector, the user association metric value for an entity corresponding to the user being determined based on a user's operational behavior for the entity.
Optionally, in one example, the apparatus may further include: and a recommending unit configured to recommend a target entity to the user based on the determined user association metric values for the respective entities.
According to another aspect of the present disclosure, there is provided a computing device comprising: at least one processor; and a memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform the method as described above.
According to another aspect of the disclosure, there is provided a non-transitory machine-readable storage medium storing executable instructions that, when executed, cause the machine to perform a method as described above.
By using the method and the device disclosed by the invention, the entity characteristic vector of each entity is updated by using the entity characteristic change value of each entity received from the neighbor user, so that the entity characteristic vector of each entity can be updated in cooperation with each user, further, when the user association metric value of each entity corresponding to each user is determined, user data does not need to be uploaded to a server, and the user association metric value of each entity corresponding to the user is determined at each user, thereby not only avoiding leakage of user privacy, but also saving data storage overhead and data operation overhead.
By using the method and the device, when the operation behavior of the user for the entity is obtained, the user preference characteristic change value and the second entity characteristic change value are determined, and then the user preference characteristic vector and the entity characteristic vector are updated based on the determined user preference characteristic change value and the second entity characteristic change value, so that the known user association metric value of the entity existing at the user can be utilized to predict the unknown entity user association metric value.
By using the method and the device disclosed by the invention, the determined second entity characteristic change value and/or the received first entity characteristic change value are transmitted to the neighbor users, so that each user can be further cooperated to predict the user association metric value of each entity, and further the decentralized recommendation system is realized.
By using the method and the device disclosed by the invention, the received first entity characteristic change value is continuously transmitted to the neighbor users based on the continuous transmission instruction, so that the transmission range of the entity characteristic change value can be controlled, each user forms a neighbor user group, and the transmission of the entity characteristic change value is transmitted in each neighbor user group, thereby carrying out user recommendation in groups and enabling the user recommendation process to be more refined.
By using the method and the device disclosed by the invention, the hash value vector is generally used for representing the user preference feature vector and the entity feature vector, so that real number operation is not needed when the user association measurement value is determined according to the user preference feature vector and the entity feature vector, thereby reducing the operation amount and saving the operation resource.
By using the method and the device, the hash vector is configured into the binary vector, so that not only can the low operation amount be further reduced, but also the data storage amount can be further reduced, and the storage resource can be saved.
Drawings
A further understanding of the nature and advantages of the present disclosure may be realized by reference to the following drawings. In the drawings, similar components or features may have the same reference numerals. The accompanying drawings are included to provide a further understanding of embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and together with the description serve to explain, without limitation, the embodiments of the disclosure. In the drawings:
FIG. 1 is a schematic diagram of a recommendation system in the prior art;
FIG. 2 is a schematic diagram of one example of a recommendation system using a method and apparatus for determining user-associated metric values for entities in accordance with embodiments of the present disclosure;
FIG. 3 is a flow chart of a method for determining a user association metric value for an entity according to one embodiment of the present disclosure;
FIG. 4 is a flow chart of a method for determining a user association metric value for an entity according to another embodiment of the present disclosure;
FIG. 5 is a schematic diagram of one example of a delivery process in a method for determining user-associated metric values for an entity in accordance with one embodiment of the present disclosure;
FIG. 6 is a block diagram of an apparatus for determining a user-associated metric value for an entity according to one embodiment of the present disclosure;
FIG. 7 is a block diagram of an apparatus for determining a user-associated metric value for an entity according to one embodiment of the present disclosure;
fig. 8 is a block diagram of a computing device for implementing a method for determining user association metric values for an entity, according to one embodiment of the present disclosure.
Detailed Description
The subject matter described herein will be discussed below with reference to example embodiments. It should be appreciated that these embodiments are discussed only to enable a person skilled in the art to better understand and thereby practice the subject matter described herein, and are not limiting of the scope, applicability, or examples set forth in the claims. Changes may be made in the function and arrangement of elements discussed without departing from the scope of the disclosure. Various examples may omit, replace, or add various procedures or components as desired. In addition, features described with respect to some examples may be combined in other examples as well.
As used herein, the term "comprising" and variations thereof mean open-ended terms, meaning "including, but not limited to. The term "based on" means "based at least in part on". The terms "one embodiment" and "an embodiment" mean "at least one embodiment. The term "another embodiment" means "at least one other embodiment". The terms "first," "second," and the like, may refer to different or the same object. Other definitions, whether explicit or implicit, may be included below. Unless the context clearly indicates otherwise, the definition of a term is consistent throughout this specification.
Methods and apparatus of the present disclosure for determining user association metric values for entities are now described with reference to the accompanying drawings.
Before explaining the method and apparatus for determining a user association metric value for an entity of the present disclosure (hereinafter referred to as a user association metric value determination method and apparatus), a recommendation system to which the user association metric value determination method and apparatus of the present disclosure is applied will be first described.
FIG. 2 is a schematic diagram of one example of a recommendation system using a user-associated metric value determination method and apparatus in accordance with embodiments of the present disclosure. As shown in fig. 2, the recommendation system 200 does not need a configuration server, and the operation behavior data 220 of each user 210 is stored at each user 210 without being transmitted outward, so that the privacy of the user can be protected. Training of the scoring model 230 is also performed at each user 210 based on the respective operational behavior data 220, and thus storage of data and training of the scoring model can be distributed to each user, thereby enabling reduction of data storage overhead and computation overhead.
To implement a recommendation system as shown in fig. 2, a plurality of users located in the recommendation system need to be cooperated to determine a user association metric value between a certain user and each entity. In the present disclosure, a plurality of users participating in collaboration are referred to as a collaborative user group, and a plurality of collaborative user groups may exist in one recommendation system. The collaborative user group may be determined based on a distance between individual users. For example, GPS information of the user terminal device may be collected, and the distance between the respective users may be determined based on the GPS information. In addition, the collaborative user group may also be determined based on the similarity between the individual users.
Fig. 3 is a flowchart of a method for determining a user association metric value for an entity according to one embodiment of the present disclosure.
As shown in FIG. 3, at block 310, the user's operational behavior for various entities is obtained. Then, at block 320, a determination is made as to whether the user's operational behavior for each entity has been obtained. The entity may be an item, a user, or information, etc. The items may be, for example, merchandise for sale on a shopping website, music on a music application, etc. The users may be, for example, individual users within a recommendation system or users in a social application, etc. The information may be, for example, various news information on an information web site. The action may be, for example, clicking, purchasing, collecting, deleting, scoring, etc.
When the operational behavior of the user for each entity is obtained, at block 330, a user preference characteristic change value for the user and a second entity characteristic change value for the entity are determined.
The user preference feature vector represents a preference of the user. Taking music as an example, if music is classified into classical music, popular music, western music, chinese music, etc., the user preference feature vector may represent the degree of preference of the user for the above-described respective categories of music. The entity feature vector represents the attributes of the entity itself. For example, taking the above respective music categories as an example, the entity feature vector may represent the likelihood that a particular music belongs to the above respective music categories.
The degree of association between a user and an entity is related to the user preferences of the user and the entity properties of the entity. For example, if a certain user has a higher preference for classical music while a certain music has a higher likelihood of belonging to classical music, it can be predicted that the degree of association of the user with the music is higher. In the present disclosure, the degree of association between a particular user and each entity is represented by a user association metric value for each entity. The user-associated metric value of an entity may be, for example, a score of the entity by the user. The user association metric value may be determined from a user preference feature vector of the user and entity feature vectors of the respective entities. In the initial state, the user preference feature vector of the user and the entity feature vector of each entity may be randomly initialized, or the user preference feature vector and the entity feature vector in the history data.
The operation behavior of the user aiming at the entity can reflect the association degree of the user and the entity. For example, if a user purchases or collects a certain commodity, it indicates that the user has a higher preference for that commodity. If the user masks a piece of information, it indicates that the user's preference for that piece of information is very low. Thus, the operational behavior of a user for an entity can reflect the user-associated metric value for that entity for that user.
Because each operation behavior can represent the association degree between the user and the entity, the user association metric value corresponding to various operation behaviors of the user aiming at the entity can be preset. For example, the user association metric corresponding to the purchase behavior of the user to the entity may be set to 1, the user association metric corresponding to the collection behavior of the user to the entity may be set to 0.8, and the user association metric corresponding to the deletion or shielding behavior of the user to the entity may be set to 0. The user association metric values corresponding to various operation behaviors can be manually summarized according to experience, and can also be predicted by adopting a machine learning method according to historical behavior data.
When the operational behaviour of a user for at least one entity is obtained, the user-associated metric value for that entity is known for that user. At this time, a user preference feature variation value with respect to an existing user preference feature vector (a randomly initialized or historical user preference feature vector) and a second entity feature variation value with respect to an existing entity feature vector may be determined from the known user association metric values.
In one example, a user preference feature change value for a user and a second entity feature change value for each entity may be determined based on a loss function. The penalty function is a function of the association metric value for the entity corresponding to the user, the user preference feature vector, and the entity feature vector. In one example, the user preference feature vector and the entity feature vector may be hash value vectors. The hash value vector may be a binary vector, such as a 0-1 value vector, -1-1 value vector, or the like. The loss function can be expressed, for example, by the following equation one.
Mathematical formula one:
wherein U is i User preference feature vector, V, representing user i j Entity feature vector representing entity j present at user i, K representing the dimension of the vector, R ij Representing a user association metric value for entity j corresponding to user i. At the number ofIn the first expression, the first term constrains the relationship between the user association metric value and the preference feature vector, i.e., the higher the user association metric value, the more similar the user preference feature vector and the entity feature vector should be. The second term controls the degree of uniformity of hash coding in the user preference feature vector as well as the entity feature vector, i.e. the ratio of the two values (e.g. -1 and 1) encoded should be as identical as possible. The parameter lambda controls the degree of this constraint, the greater lambda the stronger the constraint.
In generating the user preference feature vector and the entity feature vector represented by binary hash value vectors, the preference feature vector may first be relaxed to a real number field, e.g., between-1-1. After the preferred feature vector is found, real numbers in the preferred feature vector may be converted to hash values based on a predetermined threshold, for example, real numbers greater than 0 may be encoded as 1 and real numbers not exceeding 0 may be encoded as-1, thereby obtaining a user preferred feature vector and an entity feature vector represented by binary hash value vectors. The hash value in the hash value vector may also be determined using other hash encoding algorithms, the above being just one example of hash encoding.
In this example, the user preference feature variation value may be determined by biasing the loss function L against the user preference feature vector and the second entity feature variation value may be determined by biasing the loss function L against the entity feature vector.
After determining the user preference feature variation value and the second entity feature variation value, at block 340, the user preference feature vector of the user is updated based on the user preference feature variation value. For example, the user preference feature vector may be updated by the following equation two.
Mathematical formula two:
wherein α is the learning rate.
In addition, at block 380, the entity characteristic vector for the corresponding entity may also be updated based on the determined second entity characteristic preference change value. For example, the entity feature vector may be updated by the following equation three.
Mathematical formula three:
wherein α is the learning rate. The learning rates in the second and third expressions may take the same value or may take different values.
The second entity characteristic change value determined at block 330 may be communicated to at least one neighbor user of the user at block 350 for updating the entity characteristic vector of the corresponding entity present at the neighbor user. The neighbor users are determined from a collaborative user group including the user.
The entity characteristic change value used to update the entity characteristic vector of the corresponding entity present at the user may also be obtained from a neighboring user of the user. As shown in fig. 3, at block 360, for each of at least one entity, a first entity characteristic change value for that entity is obtained from a neighbor user of the user, and then at block 370, a determination is made as to whether the first entity characteristic preference characteristic change value has been obtained. The first entity characteristic preference characteristic variation value may be determined by a neighbor user according to the determination procedure described above.
When the first entity characteristic change value is obtained, at block 380, the entity characteristic vector of the corresponding entity may be updated based on the obtained first entity characteristic change value. In addition, the obtained first entity characteristic change value may be further communicated to at least one neighbor user of the user for updating the entity characteristic vector of the corresponding entity present at the neighbor user at block 350.
For each entity present at the user, if the user performed an operational action on the entity and received a first entity characteristic change value from a neighboring user, the entity characteristic vector for the entity may be updated based on the received first entity characteristic change value and the determined second entity characteristic change value at block 380.
After the update process, at block 390, user-associated metric values for each entity are determined based on the updated user preference feature vector and the updated entity feature vector for each entity. The user-associated metric value can be determined by, for example, the following equation four.
Mathematical formula four:
R ij ＝U i ×V j
when the hash value vector is used for representing the user preference feature vector and the entity feature vector, the user association metric value can be determined through exclusive or operation among the hash values, so that real operation is not needed when the user association metric value is determined. Taking a binary hash value vector as an example, if the dimension K of the preference feature vector is 3, it is assumed that the user preference feature vector of a certain user is [ -1,1], the entity feature vector of entity a is [ -1,1], the entity feature vector of entity B is [ -1,1], and the entity feature vector of entity C is [1, -1]. The preference feature vector of entity a is most similar to the preference feature vector of the user, so that when making a recommendation, the item most similar to the user (i.e., a) can be recommended to the user. Thus, when using hash value vectors, real vector operations need not be performed, or ordering need not be performed, but rather only by looking up hash tables to determine the most similar entities.
In addition, when the binary hash value vector is used for representing the user preference feature vector and the entity feature vector, the storage space can be saved.
In practice, the number of entities is often very large, and it is not possible for a user to operate on each entity. Thus, for entities that the user has not operated on, their user-associated metric values are unknown or can only be determined based on the unexplored entity feature vectors. Through the above-described transfer procedure, for an entity that the user has not operated, the actual entity feature vector can be updated based on the first entity feature change values transferred from the neighbor users. The first entity characteristic change value and/or the received second entity characteristic change value of the entity determined based on the operational behavior of the user may also be communicated to the neighbor user for updating the entity characteristic vector of the corresponding entity at the neighbor user.
Thus, the entity feature vectors of the respective entities existing at the respective users can be updated in cooperation with the plurality of users for the purpose of accurately determining the user association metric values of the respective entities corresponding to each user.
Fig. 4 is a flow chart of a method for determining a user association metric value for an entity according to another embodiment of the present disclosure.
As shown in fig. 4, at block 410, for each entity present at the user, the entity characteristic vector for that entity is updated based on a first entity characteristic change value for that entity, the first entity characteristic change value received from a neighbor user of the user. The first entity characteristic change value may be determined at the neighbor user according to the process of determining the second entity characteristic change value as described above.
Upon receiving the first entity characteristic change value, at block 420, a user association metric value for each entity is determined based on the user preference feature vector of the user and the updated entity feature vector of each entity.
With this embodiment, even if the user does not perform an operation for a certain entity, the entity feature vector of the corresponding entity may be updated based on the first entity feature preference change value transmitted from the neighboring user, so as to determine the user association metric value of the entity.
In one example, neighbor users may be determined from a collaborative user group that includes individual users based on their geographic location information (e.g., GPS location information). The user equipment of each user can acquire the GPS information sent by the user equipment of other users, and then the users in the preset distance range of the user in the corresponding collaborative user group can be determined to be neighbor users. By determining the neighbor users based on the distance range, the neighbor user group can be established without acquiring information of other users, and further the determination of the user association metric value is realized in cooperation with the neighbor user group.
Further, in another example, the user may be considered an entity, and the target entity may be recommended to the user as a neighbor user of the user based on the determined user association metric values for each of the at least one entity. For example, a target user with a higher degree of user association may be recommended to the user. By recommending the target user as a neighbor user based on the user association metric value, it is possible to update the entity feature vector of each entity by using a user with a high degree of association (e.g., high similarity) as the same neighbor user group.
After determining the neighbor users, a process of transferring the entity characteristic change values may be performed between the users and the neighbor users.
Fig. 5 is a schematic diagram of one example of a transfer process in a user association metric value determination method according to one embodiment of the present disclosure.
As shown in fig. 5, the physical characteristic change value may be transmitted in a plurality of degrees, and in fig. 5, a solid arrow shows a one-degree transmission process and a dotted arrow shows a two-degree transmission process. The multi-degree transmission refers to that after a user transmits the determined entity characteristic change value to a neighbor user, the neighbor user continuously transmits the received entity characteristic change value to the neighbor user. In fig. 5, user i 0 To user i 1 、i 2 、i 3 The transfer process of (1) is one-time transfer, user i 1 Continue to user i 4 The transfer process of (2) is a second degree transfer.
The multi-degree transfer may be implemented based on a continue transfer instruction. The continue delivery indication may be received from a neighbor user or may be determined based on a delivery number constraint of the delivered entity characteristic change value. For example, user i 0 Transmitting the determined entity characteristic change value to a user i 1 User i can be presented with 1 The transfer continues to transfer the indication. If user i 1 Upon receipt of the slave user i 0 When the transmitted entity characteristic change value also receives a continuous transmission instruction, the user i 1 Will continue to pass the received entity characteristic change value to its neighbor user i 4 . If user i 0 The sent continuous transmission instruction is larger than the second-degree transmission, and the user i 1 Can also continue to user i 4 The transfer continues to transfer the indication. User i 4 Upon receipt of the slave user i 1 The transfer process may continue when the transferred continuing transfer instruction and the entity characteristic change value.
User i 0 When the determined entity characteristic change value is transferred to the neighbor user, the entity characteristic change value transfer frequency indication can be transferred to the neighbor user, and the transfer frequency indication is used for restraining the user i 0 The degree of transfer (i.e., the range of transfer) of the determined physical characteristic change value. For example, when user i 0 Neighbor user i of (2) 2 Or i 3 Upon receipt of the transfer number indication, a determination may be made as to whether to proceed to i based on the transfer number indication 2 Or i 3 Is a neighbor user transfer continuation transfer instruction.
In multi-degree delivery, the entity characteristic change value delivered by the multi-degree may be given a time stamp. The timestamp identifies the time at which the entity characteristic change value was issued. Thus, when there are neighbor users common to a plurality of users in the collaborative user group, it can be determined whether the received entity characteristic change value has been previously received based on the time stamp, and it can be determined whether the received entity characteristic change value is valid based on the time stamp.
For example, when a user receives a first entity characteristic change value, if it is determined based on the time stamp that the received first entity characteristic change value is sent before the last received first characteristic change value, the received first characteristic change value is discarded. If the time stamp of the received first entity characteristic change value is later than the time stamp of the last received first entity characteristic change value, updating the corresponding entity characteristic vector at the user by using the received first entity characteristic change value.
In one example, the communicated entity characteristic change value may also have a user identification identifying the originating user of the entity characteristic change value. Thus, it may be determined whether the received first entity characteristic change value is repeatedly received based on the user identification and the time stamp.
By continuing to transmit the instruction, a neighbor user group with a predetermined scale can be realized, and then the entity feature vector is updated in cooperation with each user in the neighbor user group. Therefore, the updating process of the entity characteristic vector can be only carried out among related users, and the determined user association metric value is further refined.
Furthermore, it should be noted that, in embodiments of the present disclosure, the user association metric value may be determined based on the updated user preference feature vector and the entity feature vector each time the user preference feature vector or the entity feature vector is updated. In another example, the user association metric value may be determined based on the updated user preference feature vector and the entity feature vector when the user preference feature vector or the entity feature vector is updated a predetermined number of times. For example, for the example shown in FIG. 4, the operations of block 420 are not performed immediately after each execution of block 410.
Fig. 6 is a block diagram of a user association metric value determination device 600 according to one embodiment of the present disclosure. As shown in fig. 6, the user association metric value determination device 600 includes an updating unit 610 and a user association metric value determination unit 620.
The updating unit 610 is configured to update, for each entity present at the user, an entity characteristic vector of the entity based on a first entity characteristic change value of the entity, the first entity characteristic change value being received from a neighboring user of the user. The neighbor users are determined from a predetermined collaborative user group including the users. The user association metric value determination unit 620 is configured to determine a user association metric value for each entity based on the user preference feature vector of the user and the updated entity feature vector of each entity.
Fig. 7 is a block diagram of a user association metric value determination device 700 according to one embodiment of the present disclosure. As shown in fig. 7, the user association metric value determination device 700 includes a feature variation value determination unit 710, an update unit 720, a user association metric value determination unit 730, a feature variation value transmission unit 740, and a recommendation unit 750.
The feature change value determining unit 710 is configured to determine a user preference feature change value of the user and a second entity feature change value of the entity when the operation behavior of the user for each entity is acquired. In one example, the feature change value determination unit 710 may be configured to determine a user preference feature change value of the user and a second entity feature change value of each entity based on the loss function. Wherein the penalty function is a function of a user-associated metric value for an entity corresponding to the user, the user preference feature vector, and the entity feature vector, the user-associated metric value for the entity corresponding to the user being determined based on the user's operational behavior for the entity.
The updating unit 720 may then update the user preference feature vector of the user based on the user preference feature variation value. The updating unit 720 may be further configured to update the entity feature vector of the entity based on the second entity feature variation value of the entity and/or the first entity feature variation value of the entity.
After updating the entity feature vector and the user preference feature vector, the user association metric value determination unit 730 may determine a user association metric value for each entity based on the updated user preference feature vector and the updated entity feature vector of each entity.
After the feature change value determining unit 710 determines the second entity feature change value of each entity, the feature change value transmitting unit 740 may transmit the determined second entity feature change value of each entity to at least one neighbor user of the user. Further, the feature change value transmitting unit 740 may be further configured to transmit the first entity feature change value of each entity acquired from the neighbor user of the user to at least one neighbor user of the user.
The neighbor users of the user may also be determined based on the location information of the respective users, and the recommendation unit 750 may recommend the target entity to the user based on the determined user association metric values for the respective entities. In this example, neighbor users are recommended to the user as entities based on user association metric values.
Embodiments of the user association metric value determination method and apparatus of the present disclosure are described above with reference to fig. 1-7. It should be appreciated that the detailed description of the method embodiments above applies equally to the apparatus embodiments. The above user-associated metric determination means may be implemented in hardware, or in software, or in a combination of hardware and software.
Fig. 8 is a block diagram of a computing device 800 for implementing a method for determining user association metric values for an entity, according to one embodiment of the present disclosure.
As shown in fig. 8, computing device 800 may include at least one processor 810, a memory 820, a memory 830, a communication interface 840, and an internal bus 850, the at least one processor 810 executing at least one computer readable instruction (i.e., the elements implemented in software as described above) stored or encoded in a computer readable storage medium (i.e., memory 820).
In one embodiment, stored in memory 820 are computer-executable instructions that, when executed, cause at least one processor 810 to: updating, for each entity present at a user, an entity feature vector for the entity based on a first entity feature variation value for the entity, the first entity feature variation value received from a neighbor user of the user; and determining a user association metric value for each entity based on the user preference feature vector and the updated entity feature vector for each entity.
It should be understood that the computer-executable instructions stored in memory 820, when executed, cause at least one processor 810 to perform the various operations and functions described above in connection with fig. 1-7 in various embodiments of the present disclosure.
In this disclosure, computing device 800 may include, but is not limited to: personal computers, server computers, workstations, desktop computers, laptop computers, notebook computers, mobile computing devices, smart phones, tablet computers, cellular phones, personal Digital Assistants (PDAs), handsets, messaging devices, wearable computing devices, consumer electronic devices, and the like.
According to one embodiment, a program product, such as a non-transitory machine-readable medium, is provided. The non-transitory machine-readable medium may have instructions (i.e., elements implemented in software as described above) that, when executed by a machine, cause the machine to perform the various operations and functions described above in connection with fig. 1-7 in various embodiments of the disclosure.
In particular, a system or apparatus provided with a readable storage medium having stored thereon software program code implementing the functions of any of the above embodiments may be provided, and a computer or processor of the system or apparatus may be caused to read out and execute instructions stored in the readable storage medium.
In this case, the program code itself read from the readable medium may implement the functions of any of the above-described embodiments, and thus the machine-readable code and the readable storage medium storing the machine-readable code form part of the present invention.
Examples of readable storage media include floppy disks, hard disks, magneto-optical disks, optical disks (e.g., CD-ROMs, CD-R, CD-RWs, DVD-ROMs, DVD-RAMs, DVD-RWs), magnetic tapes, nonvolatile memory cards, and ROMs. Alternatively, the program code may be downloaded from a server computer or cloud by a communications network.
The detailed description set forth above in connection with the appended drawings describes exemplary embodiments, but does not represent all embodiments that may be implemented or fall within the scope of the claims. The term "exemplary" used throughout this specification means "serving as an example, instance, or illustration," and does not mean "preferred" or "advantageous over other embodiments. The detailed description includes specific details for the purpose of providing an understanding of the described technology. However, the techniques may be practiced without these specific details. In some instances, well-known structures and devices are shown in block diagram form in order to avoid obscuring the concepts of the described embodiments.
The alternative implementation manner of the embodiment of the present disclosure has been described in detail above with reference to the accompanying drawings, however, the embodiment of the present disclosure is not limited to the specific details in the foregoing implementation manner, and various simple modifications may be made to the technical solutions of the embodiment of the present disclosure within the scope of the technical concept of the embodiment of the present disclosure, and all the simple modifications belong to the protection scope of the embodiment of the present disclosure.
The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other variations without departing from the scope of the disclosure. Thus, the disclosure is not limited to the examples and designs described herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (16)
1. A method for determining a user association metric value for an entity, comprising:
updating, for each entity present at a user, an entity feature vector for the entity based on a first entity feature variation value for the entity, the first entity feature variation value being received from a neighbor user of the user, the neighbor user being determined from a predetermined set of collaborative users including the user, wherein the first entity feature variation value has a timestamp indicating a time of transmission of the first entity feature variation value, the timestamp being used to determine whether the received first entity feature variation value is valid when the user receives the first entity feature variation value, wherein the first entity feature variation value is transmitted to the user by the neighbor user by a multi-degree transfer, the multi-degree transfer being a transfer of the received entity feature variation value to the neighbor user of the neighbor user by one user, the neighbor user continuing to transfer the received entity feature variation value to the user of the neighbor user; and
Determining a user association metric value for each entity based on the user preference feature vector of the user and the updated entity feature vector of each entity,
wherein the method further comprises:
when the operation behaviors of the user for the entities are acquired, determining a user preference characteristic change value of the user and a second entity characteristic change value of the entity; and
updating the user preference feature vector of the user based on the user preference feature variation value,
updating the entity feature vector for the entity based on the first entity feature variation value for the entity includes:
updating the entity feature vector of the entity based on the second entity feature variation value of the entity and the first entity feature variation value of the entity,
determining user-associated metrics for the respective entities based on the user preference feature vector of the user and the updated entity feature vector of the respective entity comprises:
and determining a user association metric value for each entity based on the updated user preference feature vector and the updated entity feature vector of each entity.
2. The method of claim 1, further comprising:
And sending the second entity characteristic change value and/or the first entity characteristic change value to at least one neighbor user of the user.
3. The method of claim 2, wherein transmitting the first entity characteristic change value to at least one neighbor user of the user is based on a continue delivery indication.
4. A method as claimed in claim 3, wherein the continued delivery indication is determined based on a delivery times constraint of the first entity characteristic change value.
5. The method of claim 1, wherein determining the user preference characteristic change value for the user and the second entity characteristic change value for the respective entity comprises:
determining a user preference characteristic change value of the user and a second entity characteristic change value of the respective entity based on a loss function,
wherein the penalty function is a function of a user association metric value for an entity corresponding to the user, the user preference feature vector, and the entity feature vector, the user association metric value for an entity corresponding to the user being determined based on a user's operational behavior for the entity.
6. The method of any of claims 1 to 5, wherein the collaborative user group is determined based on a distance between individual users.
7. The method of any of claims 1 to 5, wherein the user preference feature vector and the entity feature vector are hash value vectors.
8. The method of claim 7, wherein the hash value vector is a binary vector.
9. The method of any of claims 1 to 5, wherein the entity comprises at least one of:
an article;
a user; and
information.
10. The method of any one of claims 1 to 5, further comprising:
recommending the target entity to the user based on the determined user association metric values for the entities.
11. An apparatus for determining a user-associated metric value for an entity, comprising:
an updating unit configured to update, for each entity present at a user, an entity feature vector of the entity based on a first entity feature variation value of the entity, the first entity feature variation value being received from a neighbor user of the user, the neighbor user being determined from a predetermined collaborative user group including the user, wherein the first entity feature variation value has a timestamp indicating a transmission time of the first entity feature variation value, the timestamp being used to determine whether the received first entity feature variation value is valid when the user receives the first entity feature variation value, wherein the first entity feature variation value is transmitted to the user by the neighbor user by a multi-degree transfer, the multi-degree transfer being a transfer of the determined entity feature variation value to the neighbor user of the user by one user, the neighbor user continuing to transfer the received entity feature variation value to the neighbor user of the neighbor user; and
A user association metric value determination unit configured to determine a user association metric value for each entity based on the user preference feature vector of the user and the updated entity feature vector of each entity,
wherein the apparatus further comprises:
a feature change value determining unit configured to determine, when an operation behavior of the user with respect to the respective entities is acquired, a user preference feature change value of the user and a second entity feature change value of the entity;
the updating unit is further configured to:
updating the user preference feature vector of the user based on the user preference feature change value; and
updating the entity feature vector of the entity based on the second entity feature variation value of the entity and the first entity feature variation value of the entity,
the user association metric value determination unit is configured to:
and determining a user association metric value for each entity based on the updated user preference feature vector and the updated entity feature vector of each entity.
12. The apparatus of claim 11, further comprising:
and the feature change value sending unit is configured to send the second entity feature change value and/or the first entity feature change value to at least one neighbor user of the user.
13. The apparatus of claim 11, wherein the feature variation value determination unit is configured to:
determining a user preference characteristic change value of the user and a second entity characteristic change value of the respective entity based on a loss function,
wherein the penalty function is a function of a user association metric value for an entity corresponding to the user, the user preference feature vector, and the entity feature vector, the user association metric value for an entity corresponding to the user being determined based on a user's operational behavior for the entity.
14. The apparatus of any of claims 11 to 13, further comprising:
and a recommending unit configured to recommend a target entity to the user based on the determined user association metric values for the respective entities.
15. A computing device, comprising:
at least one processor; and
a memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform the method of any of claims 1 to 10.
16. A non-transitory machine-readable storage medium storing executable instructions which, when executed, cause the machine to perform the method of any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910176407.2A CN110046258B (en) | 2019-03-08 | 2019-03-08 | Method and device for determining user association metric values for entities |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910176407.2A CN110046258B (en) | 2019-03-08 | 2019-03-08 | Method and device for determining user association metric values for entities |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110046258A CN110046258A (en) | 2019-07-23 |
| CN110046258B true CN110046258B (en) | 2023-07-11 |
Family
ID=67274696
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910176407.2A Active CN110046258B (en) | 2019-03-08 | 2019-03-08 | Method and device for determining user association metric values for entities |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110046258B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108520303A (en) * | 2018-03-02 | 2018-09-11 | 阿里巴巴集团控股有限公司 | A kind of recommendation system building method and device |
-
2019
- 2019-03-08 CN CN201910176407.2A patent/CN110046258B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108520303A (en) * | 2018-03-02 | 2018-09-11 | 阿里巴巴集团控股有限公司 | A kind of recommendation system building method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110046258A (en) | 2019-07-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105190595A (en) | Uniquely identifying a network-connected entity | |
| US9489638B2 (en) | Method and apparatus for propagating user preference information in a communications network | |
| CN109446171B (en) | Data processing method and device | |
| CN107634897B (en) | Group recommendation method and device | |
| Lian et al. | Analyzing location predictability on location-based social networks | |
| KR101356948B1 (en) | User interest inference method and system in sns using topics on social activities with neighbors | |
| CN110782286B (en) | Advertisement pushing method, advertisement pushing device, server and computer readable storage medium | |
| CN112633978B (en) | Method, device and equipment for building graphic neural network model and method, device and equipment for recommending commodities | |
| JP2011227721A (en) | Interest extraction device, interest extraction method, and interest extraction program | |
| WO2021077814A1 (en) | Push model optimization method and device executed by user terminal | |
| Prando et al. | Content-based Recommender System using Social Networks for Cold-start Users. | |
| CN110209921B (en) | Method and device for pushing media resource, storage medium and electronic device | |
| US20110264662A1 (en) | Context collection devices, context collection programs, and context collection methods | |
| US20170272362A1 (en) | Data communication systems and methods of operating data communication systems | |
| Jiang et al. | Using combined network information to predict mobile application usage | |
| CN114547439A (en) | Service optimization method based on big data and artificial intelligence and electronic commerce AI system | |
| CN108563648B (en) | Data display method and device, storage medium and electronic device | |
| CN108614845B (en) | Behavior estimation method and device based on media file | |
| CN110046258B (en) | Method and device for determining user association metric values for entities | |
| CN112559852A (en) | Information recommendation method and device | |
| CN110795631B (en) | Push model optimization and prediction method and device based on factorization machine | |
| JP5882243B2 (en) | User attribute estimator construction method, user attribute estimation method, apparatus, and program | |
| JP5979159B2 (en) | Recommendation condition correcting device, recommended condition correcting method, and recommended condition correcting program | |
| CN110322039B (en) | Click rate estimation method, server and computer readable storage medium | |
| JP6774917B2 (en) | Programs, devices and methods for estimating entrainment pressure within a group for a given behavior |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200925 Address after: English genus Applicant after: Advanced innovation technology Co.,Ltd. Address before: A four-storey 847 mailbox in Grand Cayman Capital Building, British Cayman Islands Applicant before: Alibaba Group Holding Ltd. Effective date of registration: 20200925 Address after: English genus Applicant after: Innovative advanced technology Co.,Ltd. Address before: English genus Applicant before: Advanced innovation technology Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |