CN111046188A - User preference degree determining method and device, electronic equipment and readable storage medium - Google Patents

User preference degree determining method and device, electronic equipment and readable storage medium Download PDF

Info

Publication number
CN111046188A
CN111046188A CN201911122654.0A CN201911122654A CN111046188A CN 111046188 A CN111046188 A CN 111046188A CN 201911122654 A CN201911122654 A CN 201911122654A CN 111046188 A CN111046188 A CN 111046188A
Authority
CN
China
Prior art keywords
node
target user
determining
vector representation
candidate
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
CN201911122654.0A
Other languages
Chinese (zh)
Inventor
张富峥
王仲远
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 Sankuai Online Technology Co Ltd
Original Assignee
Beijing Sankuai Online 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 Sankuai Online Technology Co Ltd filed Critical Beijing Sankuai Online Technology Co Ltd
Priority to CN201911122654.0A priority Critical patent/CN111046188A/en
Publication of CN111046188A publication Critical patent/CN111046188A/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
    • G06F16/30—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
    • 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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • 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
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G06Q30/0203—Market surveys; Market polls
    • 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)
  • Business, Economics & Management (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Strategic Management (AREA)
  • Theoretical Computer Science (AREA)
  • Development Economics (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Computational Linguistics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The embodiment of the application provides a method and a device for determining user preference, electronic equipment and a readable storage medium, and relates to the technical field of data processing. The method comprises the following steps: determining a node matched with a historical object of interest of a target user on a knowledge graph as an initial node, determining a node matched with a candidate object to be recommended to the target user on the knowledge graph as a candidate node, determining an associated node of at least one step length associated with the initial node on the knowledge graph, determining vector representation of the target user under the at least one step length according to the association degree between the associated node of the at least one step length and the candidate node, and determining the preference degree of the target user for the candidate object according to the vector representation of the candidate object and the vector representation of the target user under the at least one step length; by the adoption of the user preference determining method provided by the embodiment of the application, the accuracy of user preference determination can be improved, and the recommending effect of the recommending system is further improved.

Description

User preference degree determining method and device, electronic equipment and readable storage medium
Technical Field
The present embodiment relates to the field of data processing technologies, and in particular, to a method and an apparatus for determining a user preference degree, an electronic device, and a readable storage medium.
Background
With the rapid development of the internet technology, a user browses objects on the internet to become a normal state, various objects exist on the internet, and for an e-commerce platform, if the user can accurately touch the requirements of the user when browsing the objects and recommend interested objects to the user, the utilization rate of the user can be greatly improved, and further, the operation profit is improved.
Generally, the e-commerce platform determines the preference degree of a user on a candidate object by using a recommendation system, and recommends the candidate object according to the preference degree, so that the user preference degree is determined to occupy an important position in the recommendation system. In the existing recommendation system based on the preference degree, the accuracy of the preference degree determination is low, and the recommendation effect of the recommendation system is poor.
Disclosure of Invention
The embodiment of the application provides a method and a device for determining user preference, electronic equipment and a readable storage medium, so that the accuracy of determining the user preference is improved, and the recommendation effect of a recommendation system is further improved.
A first aspect of an embodiment of the present application provides a method for determining user preference, where the method includes:
determining a node matched with the historical interested object on a knowledge graph as an initial node according to the historical interested object of the target user;
determining nodes matched with the candidate objects on the knowledge graph as candidate nodes according to the candidate objects to be recommended to the target user;
determining an associated node of at least one step size associated with the initial node on the knowledge-graph;
determining the vector representation of the target user under at least one step according to the association degree between the associated node of the at least one step and the candidate node;
and determining the preference degree of the target user to the candidate object according to the vector representation of the candidate object and the vector representation of the target user under at least one step.
Optionally, the determining an associated node of at least one step size associated with the initial node on the knowledge-graph comprises:
sequentially taking K from 1 to K, and determining an associated node with the step length K associated with the initial node on the knowledge graph;
the determining the vector representation of the target user under at least one step according to the association degree between the associated node of the at least one step and the candidate node comprises:
according to the association degree between the association node with the step size of 1 and the candidate node, determining the vector representation of the target user under the step size of 1;
sequentially taking K from 2 to K, updating the candidate node to be a node corresponding to the vector representation of the target user in the K-1 step length, and determining the vector representation of the target user in the K step length according to the association degree between the associated node with the step length of K and the updated candidate node;
obtaining the final vector representation of the target user according to the vector representation of the target user under the 1 st to K th step lengths and the respective weights;
the determining the preference of the target user for the candidate object according to the vector representation of the candidate object and the vector representation of the target user under at least one step comprises:
and determining the preference degree of the target user to the candidate object according to the vector representation of the candidate object and the final vector representation of the target user.
Optionally, the determining the preference of the target user for the candidate object includes:
determining the preference degree of the target user to the candidate object according to the following formula:
Figure BDA0002275852290000021
wherein the content of the first and second substances,
Figure BDA0002275852290000022
is a function of the sigmoid and is,
Figure BDA0002275852290000023
is a vector representation of the candidate object,
Figure BDA0002275852290000024
is a vector representation of the target user in at least one step.
Optionally, the determining a vector representation of the target user in at least one step includes:
determining a vector representation of the target user at the kth step size according to the following formula:
Figure BDA0002275852290000031
wherein the content of the first and second substances,
Figure BDA0002275852290000032
is a vector representation of the target user at the kth step,
Figure BDA0002275852290000033
is a set of associated triples of step size k, (h)i,ri,ti) Is that
Figure BDA0002275852290000034
The ith associated triplet of step size k, where hiIs the starting node, r, corresponding to the ith associated tripletiIs the corresponding relationship type, t, of the ith associated tripleiIs the end node, p, corresponding to the ith associated tripletiIs the degree of association between the ith triplet of step size k and the candidate node.
Optionally, before the determining, according to the historical object of interest of the target user, a node of the historical object of interest, which is matched on the knowledge graph, as an initial node, the method further includes:
obtaining a plurality of historical object of interest samples carrying preference labels;
with a plurality of said carry preference degree labelsTraining a preset model to obtain a preference degree determination model, wherein the parameters in the preference degree determination model are training samples
Figure BDA0002275852290000035
R and α.
Optionally, the method further comprises:
obtaining vector representations of a plurality of objects to be recommended for the target user;
inputting the vector representations of the objects to be recommended into the preference degree determination model, and determining the preference degrees of the target user to the objects to be recommended;
and recommending at least one object to be recommended in the objects to be recommended to the target user according to the preference degrees of the target user to the objects to be recommended and the descending order of the preference degrees.
A second aspect of the embodiments of the present application provides an apparatus for determining user preference, where the apparatus includes:
the first determination module is used for determining a node matched with the historical object of interest on the knowledge graph as an initial node according to the historical object of interest of the target user;
the second determination module is used for determining the nodes matched with the candidate objects on the knowledge graph as candidate nodes according to the candidate objects to be recommended to the target user;
a third determination module to determine an associated node for at least one step size associated with the initial node on the knowledge-graph;
a fourth determining module, configured to determine, according to the association degree between the associated node of the at least one step and the candidate node, a vector representation of the target user in the at least one step;
and the determining module is used for determining the preference degree of the target user to the candidate object according to the vector representation of the candidate object and the vector representation of the target user under at least one step length.
Optionally, the third determining module includes:
the associated node determining submodule is used for sequentially taking K from 1 to K and determining the associated node which is associated with the initial node on the knowledge graph and has the step length of K;
the fourth determining module includes:
a first vector representation determining submodule, configured to determine, according to the association degree between the associated node with a step size of 1 and the candidate node, a vector representation of the target user in a step size of 1;
a second vector representation determining submodule, configured to take K from 2 to K in sequence, update a candidate node to a node corresponding to the vector representation of the target user in the K-1 th step length, and determine the vector representation of the target user in the K-th step length according to a degree of association between the associated node with the step length of K and the updated candidate node;
a third vector representation determining submodule, configured to obtain a final vector representation of the target user for vector representations of the target user in 1 st to K th step lengths and respective weights;
the determining module includes:
and the determining submodule is used for determining the preference degree of the target user to the candidate object according to the vector representation of the candidate object and the final vector representation of the target user.
Optionally, the determining module or the determining sub-module is further configured to:
determining the preference degree of the target user to the candidate object according to the following formula:
Figure BDA0002275852290000041
wherein the content of the first and second substances,
Figure BDA0002275852290000042
is a function of the sigmoid and is,
Figure BDA0002275852290000043
is the candidateA vector representation of the object is represented by,
Figure BDA0002275852290000044
is a vector representation of the target user in at least one step.
Optionally, the fourth determining module is further configured to:
determining a vector representation of the target user at the kth step size according to the following formula:
Figure BDA0002275852290000051
wherein the content of the first and second substances,
Figure BDA0002275852290000052
is a vector representation of the target user at the kth step,
Figure BDA0002275852290000053
is the set of associated nodes with step size k, (h)i,ri,ti) Is that
Figure BDA0002275852290000054
The ith associated node with step size k, wherein hiIs the starting node, r, corresponding to the ith associated tripletiIs the corresponding relationship type, t, of the ith associated tripleiIs the end node p corresponding to the ith associated tripleiIs the degree of association between the i-th associated node with step size k and the candidate node.
Optionally, the apparatus further comprises:
the system comprises a first obtaining module, a second obtaining module and a third obtaining module, wherein the first obtaining module is used for obtaining a plurality of historical interested object samples carrying preference labels;
a training module, configured to train a preset model by using a plurality of historical interested object samples carrying preference labels as training samples to obtain a preference determination model, where a parameter in the preference determination model is a
Figure BDA0002275852290000055
R and α.
Optionally, the apparatus further comprises:
a second obtaining module, configured to obtain vector representations of a plurality of objects to be recommended for the target user;
a fifth determining module, configured to input the vector representations of the multiple objects to be recommended into the preference degree determining model, and determine the preference degrees of the target user for the multiple objects to be recommended;
and the recommending module is used for recommending at least one object to be recommended in the objects to be recommended to the target user according to the preference degrees of the target user to the objects to be recommended and in a descending order of the preference degrees.
A third aspect of embodiments of the present application provides a computer-readable storage medium, on which a computer program is stored, which when executed by a processor, performs the steps in the method according to the first aspect of the present application.
A fourth aspect of the embodiments of the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor implements the steps of the method according to the first aspect of the present application when executed.
In the embodiment of the application, the vector representation of the knowledge graph is fused into the user preference degree determination method, so that the vector representation of the knowledge graph is combined with the preference degree determination process, an end-to-end preference degree determination method is constructed, the preference degree determination accuracy is improved, meanwhile, as the knowledge graph contains rich information of objects, including attributes of the objects and rich associated information among the objects, interest points of the users on the objects can be found by using knowledge graph expansion, and the preference degree determination accuracy is further improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the description of the embodiments of the present application will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to these drawings without inventive exercise.
Fig. 1 is a flowchart of a method for determining user preference according to an embodiment of the present application;
FIG. 2 is a schematic view of a knowledge graph proposed in an embodiment of the present application;
fig. 3 is a schematic diagram of a user preference determining apparatus according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some, but not all, embodiments of the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
In the prior art, preference determination based on a knowledge graph is an important means commonly used at present, and the knowledge graph is already used as important auxiliary knowledge to help improve the accuracy of personalized preference determination. The knowledge graph is used for assisting in improving the effect of determining the personalized preference degree, and the common method is that feature engineering is firstly carried out on users and objects in the knowledge graph, and then the features are used as features to be added into a preference degree determination model. The work firstly expresses nodes in the knowledge graph as vectors by a graph embedding method, and then adds the knowledge graph vectors corresponding to users and objects as features into the preference determination model. In such a method, the knowledge graph representation and the preference determination system in the optimization target are independent, so that the representation of the optimization knowledge graph is not optimized for the target determined by the personalized preference, and the effect of the preference determination system cannot be improved to the maximum.
Based on the defects of the prior art, the applicant provides a user preference determining method, which can integrate the vector representation of the knowledge graph into the preference determining method, uniformly optimize according to the target, improve the user preference determining effect and further obtain a better recommendation effect. The user preference determination method is described in detail below.
Referring to fig. 1, fig. 1 is a flowchart of a method for determining user preference according to an embodiment of the present application, and as shown in fig. 1, the method includes the following steps:
and step S11, determining the matched nodes of the historical interest objects on the knowledge graph as initial nodes according to the historical interest objects of the target users.
In this embodiment, a knowledge graph needs to be obtained first, and the knowledge graph may be obtained by newly creating a knowledge graph by any existing means, or by using a knowledge graph that has already been created. The knowledge graph can be a knowledge graph containing objects of interest of a user history, and the knowledge graph comprises a large number of objects, characteristics (including categories, attributes, labels and the like), relationships between the objects and the characteristics. Wherein, the historical interested object can be a historical interested commodity, a historical interested scenic spot, a historical interested service, etc.
In addition, it should be noted that the content contained in the knowledge graph may be different for different application scenarios. Therefore, for different recommended scenes, the knowledge graph can be established according to the scene characteristics, and the method is not particularly limited here.
The historical interesting object of the target user can be determined through the historical purchasing behavior of the target user, or according to the historical searching behavior of the target user, or according to the historical clicking and collecting behavior of the target user, and the historical interesting object is matched with the node on the knowledge graph.
After the target user historical object of interest is determined and the knowledge graph containing the target user historical object of interest is obtained, the node matched with the historical object of interest on the commodity knowledge graph can be determined as the initial node according to the target user historical object of interest. In a specific implementation, a part or all of the historical objects of interest of the target user in the historical time period before the current time may be determined, then each determined historical object of interest is compared with the knowledge graph, and nodes matching each determined historical object of interest are determined from the knowledge graph as initial nodes, and the number of the initial nodes is the same as the number of the historical objects of interest.
And step S12, determining the node matched with the candidate object on the knowledge graph as a candidate node according to the candidate object to be recommended to the target user.
In this embodiment, similar to the historical object of interest, the candidate object may be a candidate commodity, a candidate scenic spot, a candidate service, and the like, accordingly. Taking the historical interested object as the historical interested commodity and the candidate object as the candidate commodity as an example, selecting the commodity to be recommended to the target user, namely the candidate commodity, wherein the candidate commodity is also contained in the knowledge graph, namely the candidate commodity is also matched with the node on the knowledge graph, and determining the node matched with the candidate object on the knowledge graph as the candidate node.
Step S13, determining an associated node of at least one step size associated with the initial node on the knowledge-graph.
In this embodiment, the step size represents an edge between nodes on the knowledge graph, and two nodes are connected by several edges, which means that the two nodes are associated by several step sizes.
Referring to fig. 2, fig. 2 shows a schematic view of a knowledge graph provided by an embodiment of the present application, in fig. 2, assuming that a target user history object of interest is a and a candidate object is B, which are respectively matched to the knowledge graph, as in the figure, a is an initial node, B is a candidate node, an associated node of one step associated with the initial node a in the graph is C, an associated node of two steps associated with the initial node a is D, and associated nodes of three steps associated with the initial node a are E1, E2, E3, and E4.
After the initial node is determined on the knowledge-graph in step S11, an associated node of at least one step associated with the initial node may be determined on the knowledge-graph.
Step S14, determining a vector representation of the target user in at least one step according to the association degree between the associated node of the at least one step and the candidate node.
In this embodiment, any suitable prior art may be utilized to obtain the vector representation of each node on the knowledge graph, for example, the nodes in the knowledge graph may be represented as vectors by a graph embedding method, so that the vector representations of candidate nodes, initial nodes, and associated nodes of the initial nodes on the knowledge graph may be obtained. The initial node, the associated node of the initial node and the candidate node are all nodes on the knowledge graph, so that association exists among the nodes, and the vector representation of the target user historical interest object under at least one step length can be determined by utilizing the association relationship among the nodes, namely the association degree among the nodes.
Step S15, determining the preference of the target user for the candidate object according to the vector representation of the candidate object and the vector representation of the target user in at least one step.
After the vector representation of the candidate object and the vector representation of the target user in at least one step are obtained, the preference of the target user for the candidate object can be determined according to the vector representation of the candidate object and the vector representation of the target user in at least one step.
The above method is described in detail below with reference to preferred embodiments.
In a preferred embodiment of the present invention, the step S13 may include:
and S131, sequentially taking K from 1 to K, and determining the associated node with the step length K associated with the initial node on the knowledge graph.
In this embodiment, for each target user u, a node set with a step size k associated therewith is defined
Figure BDA0002275852290000091
Wherein
Figure BDA0002275852290000092
Is the set of all triples of the knowledge-graph,
Figure BDA0002275852290000093
the node combination that the user u has historically interested in matching the object to the knowledge graph can be regarded as the initial node combination of the user u on the graph.
In addition, for each user u, a triple set of step size k associated therewith is defined
Figure BDA0002275852290000094
Figure BDA0002275852290000095
After making the definition from above, the associated node with step k associated with the initial node can be determined therefrom.
After step S131 is executed to determine the associated node with the step k on the knowledge-graph associated with the initial node, step S14 may include:
and a substep S141, determining the vector representation of the target user under the 1 st step size according to the association degree between the associated node with the step size of 1 and the candidate node.
In this embodiment, for user u, a vector representation of candidate object v is given
Figure BDA0002275852290000096
In combination with the degree of association between the associated node and the candidate node, the user u can be represented as a node having a step size of 1 associated therewith
And a substep S142, sequentially taking K from 2 to K, updating the candidate node to the node corresponding to the vector representation of the target user in the K-1 th step length, and determining the vector representation of the target user in the K-th step length according to the association degree between the associated node with the step length of K and the updated candidate node.
In one example of this embodiment, when k is 2, the method is performed
Figure BDA0002275852290000101
Alternative to refer to candidate objects
Figure BDA0002275852290000102
Is represented by a vector of
Figure BDA0002275852290000103
It can be obtained that user u is represented by a node of step 2 associated therewith
Figure BDA0002275852290000104
In another example of this embodiment, the
Figure BDA0002275852290000105
Alternative to refer to candidate objects
Figure BDA0002275852290000106
Is represented by a vector of
Figure BDA0002275852290000107
It can be derived that user u is represented by the node associated with it with step size k
Figure BDA0002275852290000108
Alternatively, user u is represented by a node of step size k associated therewith
Figure BDA0002275852290000109
The calculation can be made by the following formula:
Figure BDA00022758522900001010
wherein the content of the first and second substances,
Figure BDA00022758522900001011
is the vector representation of the target user at the kth step, the value of k can be taken to be 1 to k,
Figure BDA00022758522900001012
is a step size ofk set of associated nodes, (h)i,ri,ti) Is that
Figure BDA00022758522900001013
The ith associated node with step size k is the ith associated triplet in the triplet with step size k, where h isiIs the starting node (also called head node) corresponding to the ith associated triplet, riIs the corresponding relationship type, t, of the ith associated tripleiIs the end node (also called tail node) corresponding to the ith associated triplet. p is a radical ofiIs the degree of association between the i-th associated node with step size k and the candidate node.
Optionally, the association degree between the i-th associated node with step size k and the candidate node may be calculated by the following formula:
Figure BDA00022758522900001014
wherein the content of the first and second substances,
Figure BDA00022758522900001015
is a vector representation of the candidate node and has a length d, R is a d x d matrix representation of an edge in the knowledge-graph,
Figure BDA00022758522900001016
a vector representation, R, representing the head node of one of the set of associated triplesiAn edge representation representing the ith associated triplet,
Figure BDA00022758522900001017
a vector representation of the head node representing the ith associated triplet.
And a substep S143, obtaining the final vector representation of the target user for the vector representation of the target user under the 1 st to K step lengths and the respective weights.
Obtaining vector representation of target user under 1 st to K steps
Figure BDA00022758522900001018
To
Figure BDA00022758522900001019
Then, can be based on
Figure BDA00022758522900001020
To
Figure BDA00022758522900001021
The final vector representation of the target user is obtained by calculation according to the respective weights and is recorded as
Figure BDA00022758522900001022
In performing steps S141-S143, a final vector representation of the target user is obtained
Figure BDA00022758522900001023
Thereafter, the step S15 may include:
step S151, determining the preference of the target user for the candidate object according to the vector representation of the candidate object and the final vector representation of the target user.
Preferably, the target user's preference for the candidate object
Figure BDA0002275852290000111
The determination can be made by the following formula:
Figure BDA0002275852290000112
wherein the content of the first and second substances,
Figure BDA0002275852290000113
is a function of the sigmoid and is,
Figure BDA0002275852290000114
is a vector representation of the candidate object,
Figure BDA0002275852290000116
is the vector representation of the target user in at least one step, i.e. the final vector representation of the target user.
In the embodiment, the vector representation of the knowledge graph is fused into the user preference degree determining method, so that the vector representation of the knowledge graph is combined with the preference degree determining process, an end-to-end preference degree determining method is constructed, the preference degree determining accuracy is improved, meanwhile, as the knowledge graph contains rich information of objects, including attributes of the objects and rich associated information among the objects, interest points of the users in the objects can be found by utilizing knowledge graph expansion, and the preference degree determining accuracy is further improved.
Of course, in the embodiment of the present invention, the parameters in the user preference degree determining method need to be trained in advance, and specifically, the following preferred embodiments are described in detail.
In a preferred embodiment of the present invention, before the step S11, the method may further include:
in step S1, a plurality of historical object of interest samples carrying preference tags are obtained.
Step S2, training a preset model by taking a plurality of historical interested object samples carrying preference degree labels as training samples to obtain a preference degree determination model, wherein the parameters in the preference degree determination model are
Figure BDA0002275852290000115
R and α.
In this embodiment, an object sample in the historical data can be given, a preference tag can be set on an interested object manually, a non-preference tag can be set on a non-interested object, and the object sample carrying the preference tag or the non-preference tag can be obtained as a training sampleParameters in the model
Figure BDA0002275852290000121
R and α, resulting in an updated preference determination model.
In the embodiment, the preference label or the non-preference label object of the user to the object in the historical data is used as a training sample, the end-to-end preference degree determination determining model is trained, the vector representation of the knowledge graph is fused into the recommendation model in the training process, the knowledge graph and the recommendation model are learned uniformly, and the preference degree determination effect can be effectively improved. Meanwhile, the end-to-end model reduces the complexity of the system and the management cost of the intermediate module, and is easy to maintain and continuously update.
In the embodiment of the present invention, after the preference degree determining model is obtained, the preference degree determining model may be applied to perform object recommendation, and specifically, the following steps may be included:
step S21, obtaining vector representations of a plurality of objects to be recommended for the target user.
Step S22, inputting the vector representation of the objects to be recommended into the preference degree determination model, and determining the preference degree of the target user to the objects to be recommended.
Step S23, recommending at least one object to be recommended among the objects to be recommended to the target user according to the preference degrees of the target user for the objects to be recommended and in a descending order of the preference degrees.
When objects are recommended to a target user, a plurality of objects to be recommended are generally available, nodes in a knowledge graph are represented as vectors by a graph embedding method, then the plurality of objects to be recommended are matched to the knowledge graph to obtain vector representations of the objects to be recommended, the vector representations of the plurality of objects to be recommended are input into a preference degree determination model, the preference degrees of the target user on the plurality of objects to be recommended can be determined, and after the preference degrees are obtained, the objects to be recommended which are ranked in the front are selected and recommended to the target user according to a preference degree descending principle. The object recommendation is performed by applying the preference degree determination model of the embodiment, so that the recommendation effect can be improved.
Based on the same inventive concept, an embodiment of the present application provides a user preference determination apparatus. Referring to fig. 3, fig. 3 is a schematic diagram of a user preference determining apparatus 30 according to an embodiment of the present application. As shown in fig. 3, the apparatus includes:
the first determining module 31 is configured to determine, according to a historical object of interest of a target user, a node of the historical object of interest, which is matched on a knowledge graph, as an initial node;
a second determining module 32, configured to determine, according to a candidate object to be recommended to the target user, a node of the candidate object, which is matched on the knowledge graph, as a candidate node;
a third determining module 33 for determining an associated node of at least one step size associated with the initial node on the knowledge-graph;
a fourth determining module 34, configured to determine, according to the association degree between the associated node of the at least one step and the candidate node, a vector representation of the target user in the at least one step;
a determining module 35, configured to determine, according to the vector representation of the candidate object and the vector representation of the target user in at least one step, a preference degree of the target user for the candidate object.
Optionally, the third determining module includes:
the associated node determining submodule is used for sequentially taking K from 1 to K and determining the associated node which is associated with the initial node on the knowledge graph and has the step length of K;
the fourth determining module includes:
a first vector representation determining submodule, configured to determine, according to the association degree between the associated node with a step size of 1 and the candidate node, a vector representation of the target user in a step size of 1;
a second vector representation determining submodule, configured to take K from 2 to K in sequence, update a candidate node to a node corresponding to the vector representation of the target user in the K-1 th step length, and determine the vector representation of the target user in the K-th step length according to a degree of association between the associated node with the step length of K and the updated candidate node;
a third vector representation determining submodule, configured to obtain a final vector representation of the target user for vector representations of the target user in 1 st to K th step lengths and respective weights;
the determining module includes:
and the determining submodule is used for determining the preference degree of the target user to the candidate object according to the vector representation of the candidate object and the final vector representation of the target user.
Optionally, the determining module or the determining sub-module is further configured to:
determining the preference degree of the target user to the candidate object according to the following formula:
Figure BDA0002275852290000131
wherein the content of the first and second substances,
Figure BDA0002275852290000132
is a function of the sigmoid and is,
Figure BDA0002275852290000133
is a vector representation of the candidate object,
Figure BDA0002275852290000134
is a vector representation of the target user in at least one step.
Optionally, the fourth determining module is further configured to:
determining a vector representation of the target user at the kth step size according to the following formula:
Figure BDA0002275852290000141
wherein the content of the first and second substances,
Figure BDA0002275852290000142
is a vector representation of the target user at the kth step,
Figure BDA0002275852290000143
is the set of associated nodes with step size k, (h)i,ri,ti) Is that
Figure BDA0002275852290000144
The ith associated node with step size k, wherein hiIs the starting node, r, corresponding to the ith associated tripletiIs the corresponding relationship type, t, of the ith associated tripleiIs the end node p corresponding to the ith associated tripleiIs the degree of association between the i-th associated node with step size k and the candidate node.
Optionally, the apparatus further comprises:
the system comprises a first obtaining module, a second obtaining module and a third obtaining module, wherein the first obtaining module is used for obtaining a plurality of historical interested object samples carrying preference labels;
a training module, configured to train a preset model by using a plurality of historical interested object samples carrying preference labels as training samples to obtain a preference determination model, where a parameter in the preference determination model is a
Figure BDA0002275852290000145
R and α.
Optionally, the apparatus further comprises:
a second obtaining module, configured to obtain vector representations of a plurality of objects to be recommended for the target user;
a fifth determining module, configured to input the vector representations of the multiple objects to be recommended into the preference degree determining model, and determine the preference degrees of the target user for the multiple objects to be recommended;
and the recommending module is used for recommending at least one object to be recommended in the objects to be recommended to the target user according to the preference degrees of the target user to the objects to be recommended and in a descending order of the preference degrees.
Based on the same inventive concept, another embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, which when executed by a processor implements the steps in the method according to any of the above-mentioned embodiments of the present application.
Based on the same inventive concept, another embodiment of the present application provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, the electronic device implements the steps of the method according to any of the above embodiments of the present application.
For the device embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, refer to the partial description of the method embodiment.
The embodiments in the present specification are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
As will be appreciated by one of skill in the art, embodiments of the present application may be provided as a method, apparatus, or computer program product. Accordingly, embodiments of the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
Embodiments of the present application are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing terminal to cause a series of operational steps to be performed on the computer or other programmable terminal to produce a computer implemented process such that the instructions which execute on the computer or other programmable terminal provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present application have been described, additional variations and modifications of these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including the preferred embodiment and all such alterations and modifications as fall within the true scope of the embodiments of the application.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in the process, method, object, or terminal device that comprises the element.
The method, the device, the storage medium and the electronic device for determining the user preference degree provided by the application are introduced in detail, and a specific example is applied in the text to explain the principle and the implementation of the application, and the description of the embodiment is only used for helping to understand the method and the core idea of the application; meanwhile, for a person skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.

Claims (10)

1. A method for determining user preferences, the method comprising:
determining a node matched with the historical interested object on a knowledge graph as an initial node according to the historical interested object of the target user;
determining nodes matched with the candidate objects on the knowledge graph as candidate nodes according to the candidate objects to be recommended to the target user;
determining an associated node of at least one step size associated with the initial node on the knowledge-graph;
determining the vector representation of the target user under at least one step according to the association degree between the associated node of the at least one step and the candidate node;
and determining the preference degree of the target user to the candidate object according to the vector representation of the candidate object and the vector representation of the target user under at least one step.
2. The method of claim 1, wherein determining the associated node for the at least one step size associated with the initial node on the knowledge-graph comprises:
sequentially taking K from 1 to K, and determining an associated node with the step length K associated with the initial node on the knowledge graph;
the determining the vector representation of the target user under at least one step according to the association degree between the associated node of the at least one step and the candidate node comprises:
according to the association degree between the association node with the step size of 1 and the candidate node, determining the vector representation of the target user under the step size of 1;
sequentially taking K from 2 to K, updating the candidate node to be a node corresponding to the vector representation of the target user in the K-1 step length, and determining the vector representation of the target user in the K step length according to the association degree between the associated node with the step length of K and the updated candidate node;
obtaining the final vector representation of the target user according to the vector representation of the target user under the 1 st to K th step lengths and the respective weights;
the determining the preference of the target user for the candidate object according to the vector representation of the candidate object and the vector representation of the target user under at least one step comprises:
and determining the preference degree of the target user to the candidate object according to the vector representation of the candidate object and the final vector representation of the target user.
3. The method of claim 1 or 2, wherein the determining the target user's preference for the candidate object comprises:
determining the preference degree of the target user to the candidate object according to the following formula:
Figure FDA0002275852280000021
wherein the content of the first and second substances,
Figure FDA0002275852280000022
is a function of the sigmoid and is,
Figure FDA0002275852280000023
is a vector representation of the candidate object,
Figure FDA0002275852280000024
is a vector representation of the target user in at least one step.
4. The method of claim 1, wherein the determining the vector representation of the target user in at least one step comprises:
determining a vector representation of the target user at the kth step size according to the following formula:
Figure FDA0002275852280000025
wherein the content of the first and second substances,
Figure FDA0002275852280000026
is a vector representation of the target user at the kth step,
Figure FDA0002275852280000027
is the set of associated nodes with step size k, (h)i,ri,ti) Is that
Figure FDA0002275852280000028
The ith associated node with step size k, wherein hiIs the starting node, r, corresponding to the ith associated tripletiIs the corresponding relationship type, t, of the ith associated tripleiIs the end node, p, corresponding to the ith associated tripletiIs the degree of association between the i-th associated node with step size k and the candidate node.
5. The method of claim 4, wherein before determining the node on the knowledge-graph that matches the historical object of interest based on the historical object of interest of the target user as the initial node, the method further comprises:
obtaining a plurality of historical object of interest samples carrying preference labels;
training a preset model by taking a plurality of historical interested object samples carrying preference labels as training samples to obtain a preference determination model, wherein parameters in the preference determination model are
Figure FDA0002275852280000029
R and α.
6. The method of claim 5, further comprising:
obtaining vector representations of a plurality of objects to be recommended for the target user;
inputting the vector representations of the objects to be recommended into the preference degree determination model, and determining the preference degrees of the target user to the objects to be recommended;
and recommending at least one object to be recommended in the objects to be recommended to the target user according to the preference degrees of the target user to the objects to be recommended and the descending order of the preference degrees.
7. An apparatus for determining user preference, the apparatus comprising:
the first determination module is used for determining a node matched with the historical object of interest on the knowledge graph as an initial node according to the historical object of interest of the target user;
the second determination module is used for determining the nodes matched with the candidate objects on the knowledge graph as candidate nodes according to the candidate objects to be recommended to the target user;
a third determination module to determine an associated node for at least one step size associated with the initial node on the knowledge-graph;
a fourth determining module, configured to determine, according to the association degree between the associated node of the at least one step and the candidate node, a vector representation of the target user in the at least one step;
and the determining module is used for determining the preference degree of the target user to the candidate object according to the vector representation of the candidate object and the vector representation of the target user under at least one step length.
8. The apparatus of claim 7, wherein the third determining module comprises:
the associated node determining submodule is used for sequentially taking K from 1 to K and determining the associated node which is associated with the initial node on the knowledge graph and has the step length of K;
the fourth determining module includes:
a first vector representation determining submodule, configured to determine, according to the association degree between the associated node with a step size of 1 and the candidate node, a vector representation of the target user in a step size of 1;
a second vector representation determining submodule, configured to take K from 2 to K in sequence, update a candidate node to a node corresponding to the vector representation of the target user in the K-1 th step length, and determine the vector representation of the target user in the K-th step length according to a degree of association between the associated node with the step length of K and the updated candidate node;
a third vector representation determining submodule, configured to obtain a final vector representation of the target user for vector representations of the target user in 1 st to K th step lengths and respective weights;
the determining module includes:
and the determining submodule is used for determining the preference degree of the target user to the candidate object according to the vector representation of the candidate object and the final vector representation of the target user.
9. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 6.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executed implements the steps of the method according to any of claims 1-6.
CN201911122654.0A 2019-11-15 2019-11-15 User preference degree determining method and device, electronic equipment and readable storage medium Pending CN111046188A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911122654.0A CN111046188A (en) 2019-11-15 2019-11-15 User preference degree determining method and device, electronic equipment and readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911122654.0A CN111046188A (en) 2019-11-15 2019-11-15 User preference degree determining method and device, electronic equipment and readable storage medium

Publications (1)

Publication Number Publication Date
CN111046188A true CN111046188A (en) 2020-04-21

Family

ID=70232060

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911122654.0A Pending CN111046188A (en) 2019-11-15 2019-11-15 User preference degree determining method and device, electronic equipment and readable storage medium

Country Status (1)

Country Link
CN (1) CN111046188A (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112053210A (en) * 2020-09-11 2020-12-08 深圳市梦网视讯有限公司 Commodity community classification-based associated value propagation method, system and equipment
CN112100511A (en) * 2020-11-18 2020-12-18 浙江口碑网络技术有限公司 Preference degree data obtaining method and device and electronic equipment
CN112182386A (en) * 2020-09-29 2021-01-05 中国银联股份有限公司 Target recommendation method and device based on knowledge graph
WO2021139105A1 (en) * 2020-05-29 2021-07-15 平安科技（深圳）有限公司 Data map-based search method, smart terminal, and readable storage medium
CN113742531A (en) * 2020-05-27 2021-12-03 杭州海康威视数字技术股份有限公司 Graph recommendation method and device and electronic equipment
CN115086430A (en) * 2022-06-09 2022-09-20 中国银行股份有限公司 Interference method and device of recommendation algorithm and communication system
WO2024104336A1 (en) * 2022-11-17 2024-05-23 北京字跳网络技术有限公司 Information acquisition method and apparatus, storage medium and electronic device

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113742531A (en) * 2020-05-27 2021-12-03 杭州海康威视数字技术股份有限公司 Graph recommendation method and device and electronic equipment
CN113742531B (en) * 2020-05-27 2023-09-01 杭州海康威视数字技术股份有限公司 Picture recommendation method and device and electronic equipment
WO2021139105A1 (en) * 2020-05-29 2021-07-15 平安科技（深圳）有限公司 Data map-based search method, smart terminal, and readable storage medium
CN112053210A (en) * 2020-09-11 2020-12-08 深圳市梦网视讯有限公司 Commodity community classification-based associated value propagation method, system and equipment
CN112053210B (en) * 2020-09-11 2023-12-05 深圳市梦网视讯有限公司 Commodity community classification-based association value propagation method, system and equipment
CN112182386A (en) * 2020-09-29 2021-01-05 中国银联股份有限公司 Target recommendation method and device based on knowledge graph
CN112182386B (en) * 2020-09-29 2023-12-05 中国银联股份有限公司 Target recommendation method and device based on knowledge graph
CN112100511A (en) * 2020-11-18 2020-12-18 浙江口碑网络技术有限公司 Preference degree data obtaining method and device and electronic equipment
CN115086430A (en) * 2022-06-09 2022-09-20 中国银行股份有限公司 Interference method and device of recommendation algorithm and communication system
WO2024104336A1 (en) * 2022-11-17 2024-05-23 北京字跳网络技术有限公司 Information acquisition method and apparatus, storage medium and electronic device

Similar Documents

Publication Publication Date Title
CN111046188A (en) User preference degree determining method and device, electronic equipment and readable storage medium
CN111209476B (en) Recommendation method, model generation method, device, medium and equipment
CN108268441B (en) Sentence similarity calculation method, device and system
CN111061946A (en) Scenario content recommendation method and device, electronic equipment and storage medium
CN112100529B (en) Search content ordering method and device, storage medium and electronic equipment
CN107463701B (en) Method and device for pushing information stream based on artificial intelligence
CN110033097B (en) Method and device for determining association relation between user and article based on multiple data fields
CN110209922A (en) Object recommendation method, apparatus, storage medium and computer equipment
CN110096617B (en) Video classification method and device, electronic equipment and computer-readable storage medium
CN111104599B (en) Method and device for outputting information
CN110532469B (en) Information recommendation method, device, equipment and storage medium
CN111400586A (en) Group display method, terminal, server, system and storage medium
CN111506820A (en) Recommendation model, method, device, equipment and storage medium
CN111125521A (en) Information recommendation method, device, equipment and storage medium
CN112199600A (en) Target object identification method and device
CN111460290A (en) Information recommendation method, device, equipment and storage medium
CN111209477A (en) Information recommendation method and device, electronic equipment and storage medium
CN112232546A (en) Recommendation probability estimation method and device, electronic equipment and storage medium
CN111680213B (en) Information recommendation method, data processing method and device
CN113742578A (en) Data recommendation method and device, electronic equipment and storage medium
CN112256957B (en) Information ordering method and device, electronic equipment and storage medium
CN110827078B (en) Information recommendation method, device, equipment and storage medium
CN110765352B (en) User interest identification method and device
CN110347916A (en) Cross-scenario item recommendation method, device, electronic equipment and storage medium
CN114925750A (en) Information recommendation method and device, computer readable storage medium and electronic equipment

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