CN111932308A - Data recommendation method, device and equipment - Google Patents
Data recommendation method, device and equipment Download PDFInfo
- Publication number
- CN111932308A CN111932308A CN202010812644.6A CN202010812644A CN111932308A CN 111932308 A CN111932308 A CN 111932308A CN 202010812644 A CN202010812644 A CN 202010812644A CN 111932308 A CN111932308 A CN 111932308A
- Authority
- CN
- China
- Prior art keywords
- objects
- target user
- target
- data
- interaction
- 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
Links
Images
Classifications
-
- 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
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0255—Targeted advertisements based on user history
-
- 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
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- 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
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0254—Targeted advertisements based on statistics
Abstract
The application provides a data recommendation method, a data recommendation device and data recommendation equipment, and relates to the technical field of artificial intelligence, wherein the method comprises the following steps: acquiring historical behavior data of a target user; determining an interaction matrix of a target user according to historical behavior data of the target user; the interaction matrix is used for representing the interaction relation between the target user and the m objects; inputting the interaction matrix into a target prediction model to obtain a prediction result set; the target prediction model is obtained by training based on the relation networks of m objects, and the relation networks of the m objects are used for representing the relation among the m objects; and determining at least one object recommended to the target user from the m objects according to the prediction result set. In the embodiment of the application, the relationship network of the m objects can contain newly appeared objects, and the relationship network of the m objects contains the relationships between the objects, so that the objects recommended to the user can be effectively screened from the m objects.
Description
Technical Field
The present application relates to the field of artificial intelligence technologies, and in particular, to a data recommendation method, apparatus, and device.
Background
With the development of big data technology and mobile internet technology, the scale of various information data is also increased. The user can be confronted with massive commodities when browsing webpages, online shopping and other operations, so that the user can hardly find interested commodities from a large number of commodities.
In the prior art, a collaborative filtering technology is usually adopted to screen out commodities recommended to a user from a large number of commodities, but if a new commodity appears when the collaborative filtering technology is adopted, the commodity does not have interactive behavior data purchased or browsed by the user, the commodity cannot be recommended, and the problem of cold start is caused. Therefore, commodities recommended to a user cannot be effectively screened from massive commodities by adopting the collaborative filtering technology in the prior art.
In view of the above problems, no effective solution has been proposed.
Disclosure of Invention
The embodiment of the application provides a data recommendation method, device and equipment, and aims to solve the problem that commodities recommended to a user cannot be effectively screened from massive commodities in the prior art.
The embodiment of the application provides a data recommendation method, which comprises the following steps: acquiring historical behavior data of a target user; determining an interaction matrix of the target user according to the historical behavior data of the target user; the interaction matrix is used for representing the interaction relation between the target user and m objects; inputting the interaction matrix into a target prediction model to obtain a prediction result set; the target prediction model is obtained by training based on a relationship network of the m objects, the relationship network of the m objects is used for representing the relationship among the m objects, and the prediction result set comprises the interaction probability of the target user and each object in the m objects; and determining at least one object recommended to the target user from the m objects according to the prediction result set.
An embodiment of the present application further provides a data recommendation device, including: the acquisition module is used for acquiring historical behavior data of a target user; the determining module is used for determining an interaction matrix of the target user according to the historical behavior data of the target user; the interaction matrix is used for representing the interaction relation between the target user and m objects; the prediction module is used for inputting the interaction matrix into a target prediction model to obtain a prediction result set; the target prediction model is obtained by training based on a relationship network of the m objects, the relationship network of the m objects is used for representing the relationship among the m objects, and the prediction result set comprises the interaction probability of the target user and each object in the m objects; and the processing module is used for determining at least one object recommended to the target user from the m objects according to the prediction result set.
The embodiment of the application also provides data recommendation equipment, which comprises a processor and a memory for storing processor executable instructions, wherein the processor executes the instructions to realize the steps of the data recommendation method.
The embodiment of the application also provides a computer readable storage medium, which stores computer instructions, and the instructions realize the steps of the data recommendation method when executed.
The embodiment of the application provides a data recommendation method, which can determine an interaction matrix of a target user by acquiring historical behavior data of the target user, wherein the interaction matrix can represent the interaction relationship between the target user and m objects. Further, the interaction matrix may be input into a target prediction model obtained by pre-training based on a relationship network of m objects, so as to obtain a prediction result set, where the prediction result set includes interaction probabilities between the target user and each object of the m objects. At least one object recommended to the target user may be determined from the m objects according to the prediction result set. Because the relationship network of the m objects can contain the newly appeared objects and the relationship network of the m objects can contain rich objects and relationships between the objects, even if a new object appears and the object has no interactive data with the user, the object can be recommended according to the information such as the relationships between the object and other objects, and the objects recommended to the user can be effectively screened from the m objects.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application, are incorporated in and constitute a part of this application, and are not intended to limit the application. In the drawings:
FIG. 1 is a schematic diagram of a data recommendation system provided in accordance with an embodiment of the present application;
FIG. 2 is a schematic diagram illustrating steps of a data recommendation method according to an embodiment of the present application;
FIG. 3 is a schematic structural diagram of a data recommendation device provided according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of a data recommendation device provided according to an embodiment of the present application.
Detailed Description
The principles and spirit of the present application will be described with reference to a number of exemplary embodiments. It should be understood that these embodiments are given solely for the purpose of enabling those skilled in the art to better understand and to practice the present application, and are not intended to limit the scope of the present application in any way. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
As will be appreciated by one skilled in the art, embodiments of the present application may be embodied as a system, apparatus, device, method or computer program product. Accordingly, the present disclosure may be embodied in the form of: entirely hardware, entirely software (including firmware, resident software, micro-code, etc.), or a combination of hardware and software.
Although the flow described below includes operations that occur in a particular order, it should be appreciated that the processes may include more or less operations that are performed sequentially or in parallel (e.g., using parallel processors or a multi-threaded environment).
In an example scenario of the present application, there is provided a target information determination system, as shown in fig. 1, which may include: the terminal device 101, the server 102, and the target user may initiate a request for data recommendation through the terminal device 101, for example, the target user opens a web page/application software to start browsing, clicks into a program such as "good item recommendation", guess you like "and the like. The server 102 may obtain historical behavior data of the target user based on a data recommendation request operation submitted by the user, determine an interaction matrix of the target user, and input the interaction matrix into the target prediction model to obtain a prediction result set, so that at least one object recommended to the target user may be determined from m objects according to the prediction result set.
The terminal device 101 may be a terminal device or software used by a user. Specifically, the terminal device may be a terminal device such as a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart watch, or other wearable devices, or may be a robot device. Of course, the terminal apparatus 101 may be software that can be run in the above terminal apparatus. For example: system applications, payment applications, browsers, wechat applets, and the like.
The server 102 may be a single server or a server cluster, and certainly, the functions of the server may also be implemented by a cloud computing technology. The server 102 may be connected to a plurality of terminal devices, or may be a server having a strong information set library, and may perform data screening based on a data recommendation request initiated by a user and an information set in the information set library. In some example scenarios, the server 102 may also recommend the filtered object to the user and present the object on the interface of the terminal device 101.
Referring to fig. 2, the present embodiment can provide a data recommendation method. The data recommendation method may be used for determining at least one object recommended to a target user from a plurality of objects based on an interaction matrix of the target user and a target prediction model. The data recommendation method may include the following steps.
S201: and acquiring historical behavior data of the target user.
In this embodiment, the historical behavior data of the target user may be obtained first. The target user may be a user who needs to recommend, the historical behavior data may be behavior data generated by the target user before a current time node, and the behavior data may be used to characterize behavior characteristics of the target user.
In one embodiment, the historical behavior data may include: objects purchased by the user, objects viewed by the user, objects collected by the user, objects added to the shopping cart by the user, etc. It is of course understood that, in some embodiments, the historical behavior data may further include: the frequency of purchasing the objects by the user, the time of browsing the objects by the user, the number of the objects collected by the user, and the like can be determined according to the actual situation, and the application does not limit the frequency.
In one embodiment, the manner of obtaining the historical behavior data of the target user may include: and inquiring from a database according to a preset path, or searching historical behavior data of a target user in a webpage according to a certain searching condition. It is understood that, the sample data set may also be obtained in other possible manners, for example, historical behavior data of the target user input to the system by a receiving person may be determined according to actual situations, which is not limited in this application.
S202: determining an interaction matrix of a target user according to historical behavior data of the target user; the interaction matrix is used for representing the interaction relation between the target user and the m objects.
In this embodiment, the interaction matrix of the target user may be determined according to the historical behavior data of the target user. The interaction matrix can be used for representing the interaction relationship between the target user and the m objects. Each element in the interaction matrix may represent an interaction relationship between the target user and one object of the m objects, and when the total number of elements in the interaction matrix is greater than the total number of objects, the remaining elements in the interaction matrix may be set to 0.
In the present embodiment, the object may refer to an object that is a target in action or thinking, or a specific object that is used by a user to accept or select. In some embodiments, the m objects may be objects to be recommended to the user, and the objects may include: merchandise links, video links, applications, applets, terms, etc. Of course, the m objects are not limited to the above examples, and other modifications are possible for those skilled in the art in light of the technical spirit of the present application, and are intended to be included within the scope of the present application as long as they achieve the same or similar functions and effects as the present application.
In this embodiment, the interaction relationship may include: the above interaction relationship is not limited to the above examples, and other modifications are possible for those skilled in the art in light of the spirit of the present application, but the scope of the present application is to be construed as limited only by the functions and effects of the present application.
In the present embodiment, m may be a positive integer greater than 0, for example, 100, 200, 30, etc., and may be determined in accordance with the actual situation, which is not limited in the present application.
S203: inputting the interaction matrix into a target prediction model to obtain a prediction result set; the target prediction model is obtained by training based on a relationship network of m objects, the relationship network of the m objects is used for representing the relationship among the m objects, and the prediction result set comprises the interaction probability of a target user and each object in the m objects.
In this embodiment, the interaction matrix of the target user may be input into a target prediction model, so that a prediction result set may be obtained. The prediction result set may include interaction probabilities of a target user and each of m objects, and the target prediction model may be obtained by pre-training based on a relationship network of the m objects, and may predict the interaction probabilities of the user and each of the m objects according to an interaction matrix of the user.
In this embodiment, the relationship network of m objects may be used to visually represent the relationship between m objects. In some embodiments, the relationship network may be implemented by using a knowledge graph, which is intended to describe various entities or concepts existing in the real world and their relationships, and which constitutes a huge semantic network graph, nodes represent entities or concepts, and edges are composed of attributes or relationships. The basic form of the triple mainly comprises (entity 1-relation-entity 2) and (entity-attribute value) and the like, based on the fact that the triple is a general representation mode of the knowledge graph. Each attribute-Attribute Value Pair (AVP) may be used to characterize an entity's intrinsic properties, while a relationship may be used to connect two entities, characterizing an association between them. For example, if the item is an entity, the price is an attribute, and 2069 is an attribute value, then item-price-2069 constitutes a sample of a (entity-attribute value) triple.
In one embodiment, the entities may be something that is distinguishable and independent, such as the m objects, the entities being the most basic elements in the knowledge-graph, and different relationships between different entities. Semantic classes (concepts) are collections of entities with the same characteristics, and concepts mainly refer to collections, categories, object types, and kinds of things, such as people and geography. Content is typically expressed as names, descriptions, interpretations, etc. of entities and semantic classes, which may be expressed in text, images, audio-video, etc. An attribute (value) may be an attribute value that points to it from an entity, with different attribute types corresponding to edges of different types of attributes, and with attribute values referring primarily to the value of an object-specified attribute. The relationship may be formulated as a function, in a relational network, a function that maps a plurality of nodes (entities, semantic classes, attribute values, etc.) to boolean values.
S204: and determining at least one object recommended to the target user from the m objects according to the prediction result set.
In this embodiment, since the prediction result set includes the interaction probability between the target user and each object of the m objects, at least one object recommended to the target user can be determined from the m objects according to the prediction result set, so that the target user can conveniently see the content that may be interested in the target user from the mass data.
In an embodiment, after at least one object recommended to a target user is determined from m objects, the determined at least one object may be displayed on a display interface of a client of the target user, specifically, may be displayed in an arrangement manner of pictures, links, and the like, and specifically, may be determined according to an actual situation, which is not limited in this application
In one embodiment, if the target user is not interested in one of the recommended objects, the target user may remove the object that is not interested in the user from the display interface of the client by clicking a removal button or by a sliding operation.
In an embodiment, since the prediction result set includes the interaction probability of the target user and each object of the m objects, at least one object recommended to the target user may be determined from the m objects according to the interaction probability values.
In an embodiment, the order presented to the target user may be determined according to the level of the interaction probability value, the upper portion or the middle portion of the page that may be ranked and the lower portion of the interaction probability value may be ranked on the middle portion or the lower portion of the page, which may be determined according to the actual situation, and this is not limited in this application.
In the present embodiment, since the main idea of the collaborative filtering technology in the prior art is to find the user 2 that is most similar to the user 1, the user 1 is recommended the object that the user 2 interacted with but the user 1 did not interact with. Under the condition that the number of objects to be recommended is very large, the number of objects which are interacted by each user history is very limited, so that an interaction matrix is very sparse, similar users are difficult to find, and further the effect of recommending by adopting a collaborative filtering technology is poor and the accuracy is low. In the embodiment, the target prediction model obtained by pre-training the relationship network based on the m objects is used for prediction, similar users do not need to be searched, and the relationship network of the m objects can contain rich information such as the relationships between the objects and the attributes of the objects, so that the objects can be accurately recommended to the target users even under the condition of sparse interaction matrix.
Furthermore, in this embodiment, even if there is a new object, it can be added to the relationship network in real time, and because the relationship network can contain rich information such as the relationship between the object and the attribute of the object, even if a new object appears and the object has no interactive data with the user, it can be recommended according to the information such as the relationship between the object and other objects and the attribute of the object, so that the object recommended to the user can be effectively screened out from m objects.
From the above description, it can be seen that the embodiments of the present application achieve the following technical effects: the interaction matrix of the target user can be determined by acquiring historical behavior data of the target user, wherein the interaction matrix can represent the interaction relationship between the target user and the m objects. Further, the interaction matrix may be input into a target prediction model obtained by pre-training based on a relationship network of m objects, so as to obtain a prediction result set, where the prediction result set includes interaction probabilities between the target user and each object of the m objects. At least one object recommended to the target user may be determined from the m objects according to the prediction result set. Because the relationship network of the m objects can contain the newly appeared objects and the relationship network of the m objects can contain rich objects and relationships between the objects, even if a new object appears and the object has no interactive data with the user, the object can be recommended according to the information such as the relationships between the object and other objects, and the objects recommended to the user can be effectively screened from the m objects.
In one embodiment, before inputting the interaction matrix into the target prediction model to obtain the prediction result set, the method may further include: and acquiring an information set of each object, wherein the information set can comprise the attribute of the object and the classification of the object. Further, a plurality of sets of target data may be extracted from the information sets of the respective objects, where each set of target data includes at least one of: the relationship among the entities, the attributes, the attribute values and the entities can construct a relationship network of m objects according to a plurality of groups of target data; the relation network of the m objects is used for representing the relation among all the objects.
In this embodiment, each of the target data sets may be triple data, such as (entity 1-relationship-entity 2) and (entity-attribute value).
In this embodiment, various nodes in the relational network may be established according to information such as the attributes of the objects and the classifications to which the objects belong, for example: the system comprises object nodes, object brand nodes, merchant nodes to which the objects belong, object price interval nodes, object primary classification nodes, object secondary classification nodes, object tertiary classification nodes and the like. In some embodiments, objects may also be ranked in other ways, e.g., four levels of classification, etc. Of course, the nodes in the above-mentioned relational network are not limited to the above-mentioned examples, and other modifications are possible for those skilled in the art in light of the technical spirit of the present application, and are intended to be included within the scope of the present application as long as the functions and effects achieved by the nodes are the same as or similar to those of the present application.
In this embodiment, the three-level classification of the object may be, for example: the clothing-men's clothing-shirt, wherein, clothing, men's clothing and shirt are respectively the first grade, second grade and third grade classification, and the object node is directly connected with the third grade classification node, then the third grade classification node is connected to the second grade classification node, and the second grade classification node is connected to the first grade classification node. Therefore, the nodes can analyze and summarize the relationship in which they directly exist, and various edge relationships in the relationship network can be established, including: the object brand relation, the object corresponding merchant relation, the object price relation, the object classification relation, the classification and classification relation, the brand and classification relation and the like. The classification and classification relationship may be a relationship between a primary classification and a secondary classification, a relationship between a secondary classification and a primary classification, and the like.
In this embodiment, the above-mentioned relationship network files related to the nodes and edges may be loaded into the graph database, and the nodes and edges from the current node may be queried by setting the starting node and the required several-hop relationship when querying in the graph database. All the initial nodes are preset, and the initial nodes can be manually set according to the nodes which the user wants to inquire when inquiring. The relationship of several hops can be used to represent the node span in the relationship network, for example, one hop is equivalent to a commodity node to a merchant node, and two hops is equivalent to a commodity node to a merchant node and then to a classification node.
In this embodiment, the graph database may be: JanusGraph, Neo4j, and the like. The janus graph is an extensible graph database that can store graphs containing hundreds of billions of vertices and edges on a multi-machine cluster. It supports transactions, supporting thousands of users accessing the graph stored therein in real-time, concurrently. Neo4j is a high-performance graph database that stores structured data on a network rather than in tables.
In one embodiment, before inputting the interaction matrix into the target prediction model to obtain the prediction result set, the method may further include: acquiring a sample data set, wherein the sample data set may include interaction matrices of n sample users. Further, a relation network of m objects can be obtained, and the sample data set and the relation network of m objects are used as input data of the model to be trained, so that a target prediction model is obtained.
In the present embodiment, n may be a positive integer greater than 0, for example, 300, 2000, 3100, or the like, and may be specifically determined according to the actual circumstances, and the present application does not limit this.
In this embodiment, the n sample users may be determined from a plurality of dimensions, and the plurality of dimensions may include, but are not limited to, at least one of: different age groups, different regions, different professions, different sexes, etc., so as to ensure the representativeness of the sample data set.
In one embodiment, the model may be a rippeenet model, which is an end-to-end framework that can simulate a process of propagating user interests through a knowledge graph. RippleNet stimulates the propagation of a user's preferences for a set of knowledge entities by automatically, iteratively extending the user's potential interests along links in the knowledge graph. Thus, a superposition of multiple "ripples" activated by objects that the user has interacted with in the past forms a distribution of the user's preferences with respect to the candidate objects, which can be used to predict the final interaction probability.
In the present embodiment, the input data of the target prediction model is an interaction matrix of the user, and the output data is a concept of interaction between the user and each object in the relationship network of m objects. In this embodiment, the relational network of m objects and the sample data set may be used as input data in the training process, and the training may be performed using a Ripple Net model.
In one embodiment, obtaining the sample data set may include: the historical behavior data of the n sample users are obtained, further, the interaction matrix of each sample user can be determined according to the historical behavior data of the n sample users, and therefore the interaction matrix of each sample user can be used as a sample data set.
In this embodiment, determining the interaction matrix of the target user according to the historical behavior data of the target user may include: and generating an initial interaction matrix according to the m objects, wherein each element in the initial interaction matrix corresponds to one object respectively. Further, setting the value of each element in the initial interaction matrix according to the historical behavior data of the target user to obtain the interaction matrix of the target user; wherein, in a case that it is determined that there is an interaction relationship between the target user and a first object of the m objects, a value of an element in an initial interaction matrix corresponding to the first object may be set to 1; in the case that it is determined that there is no interaction relationship between the target user and the second object of the m objects, the value of the element in the initial interaction matrix corresponding to the second object may be set to 0.
In this embodiment, the value of each element in the initial interaction matrix may be 0, and may also be other values, which may be set according to the actual situation, and this application is not limited to this.
In this embodiment, the set interaction matrix of the target user may be as follows:
the interactive matrix comprises 9 objects in total, and the value of the element in the 1 st row and the 1 st column in the interactive matrix is 1, which indicates that the target user has an interactive relationship with the object corresponding to the element in the first row and the first column; the value of the element in the 1 st row and the 2 nd column in the interaction matrix is 0, which indicates that the target user does not have an interaction relation with the corresponding element in the 1 st row and the 2 nd column, and so on.
In one embodiment, determining at least one object recommended to the target user from the m objects according to the prediction result set may include: and performing descending order arrangement on the interaction probability of the target user and each object in the prediction result set, and taking the preset number of objects before the ordering as objects recommended to the target user. The preset data amount may be a positive integer greater than 0, for example: 10. 15, etc., which can be determined according to practical situations, and the application is not limited to this.
In one embodiment, the target user interaction matrix can be updated in real time based on the latest behavior data of the target user, so that the object recommended to the target user is updated in real time, and the accuracy and the effectiveness of the object recommended to the target user are effectively improved.
Based on the same inventive concept, the embodiment of the present application further provides a data recommendation device, such as the following embodiments. Because the principle of solving the problems of the data recommendation device is similar to that of the data recommendation method, the implementation of the data recommendation device can refer to the implementation of the data recommendation method, and repeated parts are not described again. As used hereinafter, the term "unit" or "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated. Fig. 3 is a block diagram of a data recommendation device according to an embodiment of the present application, and as shown in fig. 3, the data recommendation device may include: the following describes the configuration of the acquisition module 301, the determination module 302, the prediction module 303, and the processing module 304.
An obtaining module 301, configured to obtain historical behavior data of a target user;
the determining module 302 may be configured to determine an interaction matrix of the target user according to historical behavior data of the target user; the interaction matrix is used for representing the interaction relation between a target user and m objects;
the prediction module 303 may be configured to input the interaction matrix into the target prediction model to obtain a prediction result set; the target prediction model is obtained by training based on a relationship network of m objects, the relationship network of the m objects is used for representing the relationship among the m objects, and the prediction result set comprises the interaction probability of a target user and each object in the m objects;
the processing module 304 may be configured to determine, according to the prediction result set, at least one object recommended to the target user from the m objects.
The embodiment of the present application further provides an electronic device, which may specifically refer to a schematic structural diagram of the electronic device based on the data recommendation method provided in the embodiment of the present application shown in fig. 4, and the electronic device may specifically include an input device 41, a processor 42, and a memory 43. The input device 41 may be specifically configured to input historical behavior data of the target user. The processor 42 may be specifically configured to obtain historical behavior data of the target user; determining an interaction matrix of a target user according to historical behavior data of the target user; the interaction matrix is used for representing the interaction relation between a target user and m objects; inputting the interaction matrix into a target prediction model to obtain a prediction result set; the target prediction model is obtained by training based on a relationship network of m objects, the relationship network of the m objects is used for representing the relationship among the m objects, and the prediction result set comprises the interaction probability of a target user and each object in the m objects; and determining at least one object recommended to the target user from the m objects according to the prediction result set. The memory 43 may be specifically configured to store parameters of an interaction matrix of a target user, at least one object recommended to the target user, and the like.
In this embodiment, the input device may be one of the main apparatuses for information exchange between a user and a computer system. The input devices may include a keyboard, mouse, camera, scanner, light pen, handwriting input panel, voice input device, etc.; the input device is used to input raw data and a program for processing the data into the computer. The input device can also acquire and receive data transmitted by other modules, units and devices. The processor may be implemented in any suitable way. For example, the processor may take the form of, for example, a microprocessor or processor and a computer-readable medium that stores computer-readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, an Application Specific Integrated Circuit (ASIC), a programmable logic controller, an embedded microcontroller, and so forth. The memory may in particular be a memory device used in modern information technology for storing information. The memory may include multiple levels, and in a digital system, memory may be used as long as binary data can be stored; in an integrated circuit, a circuit without a physical form and with a storage function is also called a memory, such as a RAM, a FIFO and the like; in the system, the storage device in physical form is also called a memory, such as a memory bank, a TF card and the like.
In this embodiment, the functions and effects specifically realized by the electronic device can be explained by comparing with other embodiments, and are not described herein again.
The embodiment of the present application further provides a computer storage medium based on a data recommendation method, where the computer storage medium stores computer program instructions, and when the computer program instructions are executed, the computer storage medium may implement: acquiring historical behavior data of a target user; determining an interaction matrix of a target user according to historical behavior data of the target user; the interaction matrix is used for representing the interaction relation between a target user and m objects; inputting the interaction matrix into a target prediction model to obtain a prediction result set; the target prediction model is obtained by training based on a relationship network of m objects, the relationship network of the m objects is used for representing the relationship among the m objects, and the prediction result set comprises the interaction probability of a target user and each object in the m objects; and determining at least one object recommended to the target user from the m objects according to the prediction result set.
In this embodiment, the storage medium includes, but is not limited to, a Random Access Memory (RAM), a Read-Only Memory (ROM), a Cache (Cache), a Hard Disk Drive (HDD), or a Memory Card (Memory Card). The memory may be used to store computer program instructions. The network communication unit may be an interface for performing network connection communication, which is set in accordance with a standard prescribed by a communication protocol.
In this embodiment, the functions and effects specifically realized by the program instructions stored in the computer storage medium can be explained by comparing with other embodiments, and are not described herein again.
It will be apparent to those skilled in the art that the modules or steps of the embodiments of the present application described above may be implemented by a general purpose computing device, they may be centralized on a single computing device or distributed across a network of multiple computing devices, and alternatively, they may be implemented by program code executable by a computing device, such that they may be stored in a storage device and executed by a computing device, and in some cases, the steps shown or described may be performed in an order different from that described herein, or they may be separately fabricated into individual integrated circuit modules, or multiple ones of them may be fabricated into a single integrated circuit module. Thus, embodiments of the present application are not limited to any specific combination of hardware and software.
Although the present application provides method steps as in the above-described embodiments or flowcharts, additional or fewer steps may be included in the method, based on conventional or non-inventive efforts. In the case of steps where no necessary causal relationship exists logically, the order of execution of the steps is not limited to that provided by the embodiments of the present application. When implemented in an actual apparatus or end product, the methods of (1) can be performed sequentially or in parallel (e.g., in the context of parallel processors or multi-threaded processing) according to the embodiments or methods shown in the figures.
It is to be understood that the above description is intended to be illustrative, and not restrictive. Many embodiments and many applications other than the examples provided will be apparent to those of skill in the art upon reading the above description. The scope of the application should, therefore, be determined not with reference to the above description, but instead should be determined with reference to the pending claims along with the full scope of equivalents to which such claims are entitled.
The above description is only a preferred embodiment of the present application and is not intended to limit the present application, and it will be apparent to those skilled in the art that various modifications and variations can be made in the embodiment of the present application. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (10)
1. A method for recommending data, comprising:
acquiring historical behavior data of a target user;
determining an interaction matrix of the target user according to the historical behavior data of the target user; the interaction matrix is used for representing the interaction relation between the target user and m objects;
inputting the interaction matrix into a target prediction model to obtain a prediction result set; the target prediction model is obtained by training based on a relationship network of the m objects, the relationship network of the m objects is used for representing the relationship among the m objects, and the prediction result set comprises the interaction probability of the target user and each object in the m objects;
and determining at least one object recommended to the target user from the m objects according to the prediction result set.
2. The method of claim 1, prior to inputting the interaction matrix into the target prediction model to obtain a set of predicted results, further comprising:
acquiring information sets of the objects; wherein the information set comprises attributes of the objects and the classifications to which the objects belong;
extracting a plurality of groups of target data from the information set of each object, wherein each group of target data comprises at least one of the following data: entities, attributes, attribute values, and relationships between entities;
constructing a relationship network of the m objects according to the multiple groups of target data; wherein the relationship network of the m objects is used for characterizing the relationship between the objects.
3. The method of claim 1, prior to inputting the interaction matrix into the target prediction model to obtain a set of predicted results, further comprising:
acquiring a sample data set; wherein the sample data set comprises interaction matrixes of n sample users;
acquiring a relationship network of the m objects;
and training by taking the sample data set and the relation network of the m objects as input data of a model to obtain the target prediction model.
4. The method of claim 3, wherein obtaining a sample data set comprises:
acquiring historical behavior data of n sample users;
determining an interaction matrix of each sample user according to the historical behavior data of the n sample users;
and taking the interaction matrix of each sample user as a sample data set.
5. The method of claim 3, wherein training the set of sample data and the relational network of m objects as input data for a model comprises:
and taking the sample data set and the relation network of the m objects as input data, and training by using a Ripple Net model.
6. The method of claim 1, wherein determining the interaction matrix of the target user based on historical behavior data of the target user comprises:
generating an initial interaction matrix according to the m objects; each element in the initial interaction matrix corresponds to an object respectively;
setting the value of each element in the initial interaction matrix according to the historical behavior data of the target user to obtain the interaction matrix of the target user; wherein in the event that it is determined that there is an interaction relationship between a target user and a first object of the m objects, setting a value of an element in an initial interaction matrix corresponding to the first object to 1; and under the condition that the target user is determined not to have the interactive relation with the second object in the m objects, setting the value of an element in the initial interactive matrix corresponding to the second object to be 0.
7. The method of claim 1, wherein determining at least one object from the m objects that is recommended to the target user based on the set of predicted results comprises:
the interaction probabilities of the target users and the objects in the prediction result set are arranged in a descending order;
and taking the preset number of objects before sorting as the objects recommended to the target user.
8. A data recommendation device, comprising:
the acquisition module is used for acquiring historical behavior data of a target user;
the determining module is used for determining an interaction matrix of the target user according to the historical behavior data of the target user; the interaction matrix is used for representing the interaction relation between the target user and m objects;
the prediction module is used for inputting the interaction matrix into a target prediction model to obtain a prediction result set; the target prediction model is obtained by training based on a relationship network of the m objects, the relationship network of the m objects is used for representing the relationship among the m objects, and the prediction result set comprises the interaction probability of the target user and each object in the m objects;
and the processing module is used for determining at least one object recommended to the target user from the m objects according to the prediction result set.
9. A data recommendation device comprising a processor and a memory for storing processor-executable instructions which, when executed by the processor, implement the steps of the method of any one of claims 1 to 7.
10. A computer-readable storage medium having stored thereon computer instructions which, when executed, implement the steps of the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010812644.6A CN111932308A (en) | 2020-08-13 | 2020-08-13 | Data recommendation method, device and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010812644.6A CN111932308A (en) | 2020-08-13 | 2020-08-13 | Data recommendation method, device and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111932308A true CN111932308A (en) | 2020-11-13 |
Family
ID=73311305
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010812644.6A Pending CN111932308A (en) | 2020-08-13 | 2020-08-13 | Data recommendation method, device and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111932308A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112269517A (en) * | 2020-11-16 | 2021-01-26 | 北京百度网讯科技有限公司 | Generation method and device of interactive interface |
| CN112464042A (en) * | 2020-11-25 | 2021-03-09 | 平安科技（深圳）有限公司 | Task label generation method according to relation graph convolution network and related device |
| CN115757979A (en) * | 2022-12-20 | 2023-03-07 | 济南亿凛呈嗒网络有限公司 | User data recommendation method and system based on artificial intelligence |
| CN117632905A (en) * | 2023-11-28 | 2024-03-01 | 广州视声智能科技有限公司 | Database management method and system based on cloud use records |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110717106A (en) * | 2019-10-14 | 2020-01-21 | 支付宝(杭州)信息技术有限公司 | Information pushing method and device |
| CN111079018A (en) * | 2019-12-19 | 2020-04-28 | 深圳中兴网信科技有限公司 | Exercise personalized recommendation method, exercise personalized recommendation device, exercise personalized recommendation equipment and computer readable storage medium |
| US10650432B1 (en) * | 2016-11-28 | 2020-05-12 | Amazon Technologies, Inc. | Recommendation system using improved neural network |
| CN111179031A (en) * | 2019-12-23 | 2020-05-19 | 第四范式（北京）技术有限公司 | Training method, device and system for commodity recommendation model |
| CN111259133A (en) * | 2020-01-17 | 2020-06-09 | 成都信息工程大学 | Personalized recommendation method integrating multiple information |
| CN111507796A (en) * | 2020-04-07 | 2020-08-07 | 桂林电子科技大学 | Online shopping mall commodity recommendation method based on knowledge graph |
| CN111522886A (en) * | 2019-01-17 | 2020-08-11 | 中国移动通信有限公司研究院 | Information recommendation method, terminal and storage medium |
-
2020
- 2020-08-13 CN CN202010812644.6A patent/CN111932308A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10650432B1 (en) * | 2016-11-28 | 2020-05-12 | Amazon Technologies, Inc. | Recommendation system using improved neural network |
| CN111522886A (en) * | 2019-01-17 | 2020-08-11 | 中国移动通信有限公司研究院 | Information recommendation method, terminal and storage medium |
| CN110717106A (en) * | 2019-10-14 | 2020-01-21 | 支付宝(杭州)信息技术有限公司 | Information pushing method and device |
| CN111079018A (en) * | 2019-12-19 | 2020-04-28 | 深圳中兴网信科技有限公司 | Exercise personalized recommendation method, exercise personalized recommendation device, exercise personalized recommendation equipment and computer readable storage medium |
| CN111179031A (en) * | 2019-12-23 | 2020-05-19 | 第四范式（北京）技术有限公司 | Training method, device and system for commodity recommendation model |
| CN111259133A (en) * | 2020-01-17 | 2020-06-09 | 成都信息工程大学 | Personalized recommendation method integrating multiple information |
| CN111507796A (en) * | 2020-04-07 | 2020-08-07 | 桂林电子科技大学 | Online shopping mall commodity recommendation method based on knowledge graph |
Non-Patent Citations (1)
| Title |
|---|
| HONGWEI WANG ET AL.: "RippleNet: Propagating User Preferences on the Knowledge Graph for Recommender Systems", CIKM\'18: PROCEEDINGS OF THE 27TH ACM INTERNATIONAL CONFERENCE ON INFORMATION AND KNOWLEDGE MANAGEMENT, pages 1 - 5 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112269517A (en) * | 2020-11-16 | 2021-01-26 | 北京百度网讯科技有限公司 | Generation method and device of interactive interface |
| CN112464042A (en) * | 2020-11-25 | 2021-03-09 | 平安科技（深圳）有限公司 | Task label generation method according to relation graph convolution network and related device |
| WO2021213156A1 (en) * | 2020-11-25 | 2021-10-28 | 平安科技（深圳）有限公司 | Method and related apparatus for generating task label on basis of relationship graph convolutional network |
| CN112464042B (en) * | 2020-11-25 | 2023-07-28 | 平安科技（深圳）有限公司 | Task label generating method and related device for convolution network according to relationship graph |
| CN115757979A (en) * | 2022-12-20 | 2023-03-07 | 济南亿凛呈嗒网络有限公司 | User data recommendation method and system based on artificial intelligence |
| CN115757979B (en) * | 2022-12-20 | 2023-12-01 | 海南达润丰企业管理合伙企业(有限合伙) | User data recommendation method and system based on artificial intelligence |
| CN117632905A (en) * | 2023-11-28 | 2024-03-01 | 广州视声智能科技有限公司 | Database management method and system based on cloud use records |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11710054B2 (en) | Information recommendation method, apparatus, and server based on user data in an online forum | |
| CN109697629B (en) | Product data pushing method and device, storage medium and computer equipment | |
| JP5717858B2 (en) | Text set matching | |
| CN111932308A (en) | Data recommendation method, device and equipment | |
| WO2015188699A1 (en) | Item recommendation method and device | |
| CN106844407B (en) | Tag network generation method and system based on data set correlation | |
| CN109471978B (en) | Electronic resource recommendation method and device | |
| CN110110201B (en) | Content recommendation method and system | |
| CN105159910A (en) | Information recommendation method and device | |
| CN109460519B (en) | Browsing object recommendation method and device, storage medium and server | |
| CN112750011A (en) | Commodity recommendation method and device and electronic equipment | |
| US11423096B2 (en) | Method and apparatus for outputting information | |
| CN111259220B (en) | Data acquisition method and system based on big data | |
| CN112052387A (en) | Content recommendation method and device and computer readable storage medium | |
| CN103514282A (en) | Method and device for displaying search results of videos | |
| CN108520007A (en) | Web page information extracting method, storage medium and computer equipment | |
| CN107315753B (en) | Paging method and device across multiple databases | |
| CN113836131A (en) | Big data cleaning method and device, computer equipment and storage medium | |
| CN112116426A (en) | Method and device for pushing article information | |
| CN111400456A (en) | Information recommendation method and device | |
| CN116823410B (en) | Data processing method, object processing method, recommending method and computing device | |
| CN116955817A (en) | Content recommendation method, device, electronic equipment and storage medium | |
| CN106575418A (en) | Suggested keywords | |
| CN107622125B (en) | Information crawling method and device and electronic equipment | |
| CN115423555A (en) | Commodity recommendation method and device, electronic equipment and storage medium |
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 |