CN112084376A - Map knowledge based recommendation method and system and electronic device - Google Patents

Abstract

The invention discloses a recommendation method, a recommendation system and an electronic device based on graph knowledge. The recommendation method based on graph knowledge includes: step S1: acquiring data provided by a user and sorting the data; step S2: extracting from the sorted data Graph elements and construct knowledge graph data based on the graph elements; step S3: perform model training according to the knowledge graph data; step S4: obtain recommendation results through the trained model according to real-time data. The present invention displays the correlation between the previous behavior and the result in the form of association relationship, and can clearly check that the relationship between some behaviors and entities is relatively close, and the close relationship can be bound in the subsequent target optimization scenarios. make relevant adjustments.

Description

Recommendation method, recommendation system and electronic device based on graph knowledge

technical field

The present invention relates to a recommendation method, a recommendation system, and an electronic device, and in particular, to a recommendation method, a recommendation system, and an electronic device based on the interpretation of knowledge graphs displayed by group users.

Background technique

With the rapid development of the network, the rapid expansion of information, and the acceleration of user behavior rhythm, there is more and more demand for information that can quickly obtain users' direct or potential needs. Under this demand, the application of recommendation systems is becoming more and more extensive.

Customers not only need to view various data indicators of the recommended location, but also expect to view the relationship in the recommendation process. Build a clear relationship display through the knowledge graph, check the rationality of the recommended content, and draw deeper insights through the analysis of the relationship.

In the prior art, in the data report corresponding to the recommendation position, data indicators of different behaviors (such as exposure, click, purchase, like, etc.) are displayed. However, in practice, there are still the following shortcomings:

1. The tables of data indicators must correspond exactly in time, and cannot show the behavior sequence characteristics and relationships. Viewing behavioral data in isolation can leave out data phenomena or draw erroneous conclusions.

2. Convert the user's own characteristics and behaviors into tags, and display the data of each dimension of a single tag or multiple tags.

3. There is no time sequence or relationship between tags and tags, and it is impossible to progress in the upper layer of the data display, resulting in the effect of downward analysis from a single dimension.

Therefore, there is an urgent need to develop a recommendation method, recommendation system and electronic device based on graph knowledge to overcome the above shortcomings.

SUMMARY OF THE INVENTION

In view of the above problems, the present invention provides a kind of recommendation method based on graph knowledge, wherein, including:

Step S1: obtain the data provided by the user and organize the data;

Step S2: extracting map elements from the sorted data and constructing knowledge map data based on the association relationship between the map elements;

Step S3: performing model training according to the knowledge graph data;

Step S4: Obtain recommendation results through the trained model according to the real-time data.

The above-mentioned recommendation method based on graph knowledge, wherein, also includes:

Step S5: output a single user association track in combination with the user's historical behavior and the recommendation result.

The above-mentioned recommendation method based on graph knowledge, wherein, also includes:

Step S6: extracting the data of each user for aggregation statistics and outputting the statistical results.

In the above-mentioned recommendation method based on graph knowledge, the data includes: user data and material data, and the user data includes user behavior and user characteristics.

In the above-mentioned recommendation method based on graph knowledge, the step S1 includes defining a label structure for the data.

The above-mentioned recommendation method based on graph knowledge, wherein, the step S2 includes:

Step S21: Process the data through word segmentation and semantic processing;

Step S22: perform entity extraction, relation extraction and event extraction;

Step S23: Corresponding labels, associations and entities to form knowledge graph data.

In the above-mentioned recommendation method based on graph knowledge, the step S5 includes selecting the historical behavior of the user referenced by the model and recording the classification of different behaviors of the user, and outputting a single user associated track according to the historical behavior, classification and the recommendation result. .

In the above-mentioned recommendation method based on graph knowledge, the step S6 includes: extracting the user behavior and user characteristics of each user, and using different characteristics as dimensions to aggregate statistics and output the statistical results.

The present invention also provides a recommendation system based on graph knowledge, including:

Data processing unit: obtain the data provided by the user and organize the data;

Construction unit: extracting map elements from the sorted data and constructing knowledge map data based on the relationship between the map elements;

Training unit: perform model training according to the knowledge map data;

Result output unit: Obtain recommendation results through the trained model according to real-time data;

Trajectory output unit: output a single user associated trajectory in combination with the user's historical behavior and recommendation results;

Statistical result output unit: extracts the data of each user for aggregate statistics and outputs the statistical results.

The present invention also provides an electronic device comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the graph-based mapping described in any one of the above by the computer program. Recommended methods of knowledge.

To sum up, compared with the prior art, the present invention has the following effects: the present invention can use the markers of entity relationships in the knowledge graph and the results of strong association relationships to summarize the behavioral characteristics of a single user, and summarize the general behavioral trajectory of group users. and behavior characteristics, using graphics to display clear associations, which is convenient for quickly discovering data phenomena and performing data analysis; at the same time, based on the present invention, because of the diversity and complexity of associations, the structure of the relationship graph can be displayed in a focused manner.

Other features and advantages of the invention will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention may be realized and attained by the structure pointed out in the description, claims and drawings.

Description of drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative efforts.

Fig. 1 is the flow chart of the recommended method of the present invention;

Fig. 2 is the flow chart of step S2 in Fig. 1;

Fig. 3 is the application schematic diagram of the recommended method of the present invention;

FIG. 4 is a diagram showing an example of the recommendation method of the present invention;

5 is a schematic structural diagram of a recommendation system of the present invention;

FIG. 6 is a schematic diagram of the hardware structure of the electronic device of the present invention.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments These are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

The illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention. In addition, elements/members with the same or similar reference numerals used in the drawings and the embodiments are intended to represent the same or similar parts.

The terms "first", "second", "S1", "S2", ... etc. used in this document do not specifically refer to the order or order, nor are they used to limit the present invention, and are only used to distinguish between Elements or operations described in the same technical terms.

Orientation terms used herein, such as: up, down, left, right, front or rear, etc., refer only to the orientation of the drawings. Therefore, the directional language used is intended to be illustrative and not intended to limit this creation.

As used herein, "comprising," "including," "having," "containing," and the like, are open-ended terms, meaning including but not limited to.

As used herein, "and/or" includes any and all combinations of the stated things.

As used herein, "plurality" includes "two" and "two or more"; as used herein, "multiple groups" includes "two groups" and "two or more groups."

As used herein, the terms "approximately", "about" and the like are used to modify any quantity or error which may vary slightly, but which do not alter its essence. In general, the range of slight variation or error modified by such terms can be 20% in some embodiments, 10% in some embodiments, 5% in some embodiments, or other values. Those skilled in the art should understand that the aforementioned values can be adjusted according to actual needs, and are not limited thereto.

Certain terms used to describe the application are discussed below or elsewhere in this specification to provide those skilled in the art with additional guidance in the description of the application.

Knowledge Graph (Knowledge Graph), known as knowledge domain visualization or knowledge domain mapping map in the library and information industry, is a series of various graphs showing the knowledge development process and structural relationship. , analyze, construct, map and display knowledge and the interconnections between them.

Knowledge graph is a combination of theories and methods of applied mathematics, graphics, information visualization technology, information science and other disciplines with quantitative citation analysis, co-occurrence analysis and other methods, and uses a visual graph to visualize the core structure and development of the discipline. Modern theories of history, frontier areas, and overall knowledge structures for the purpose of multidisciplinary integration. It can provide practical and valuable reference for disciplinary research.

Knowledge graph is a graph-based data structure, consisting of nodes (points) and edges (Edges), each node represents an "entity", and each edge is a "relationship" between entities and entities. is the semantic web. Entities can refer to things in the real world, such as people, place names, companies, telephones, animals, etc.; relationships are used to express a certain connection between different entities. From the above picture, we can see that entities have place names and people; Dali belongs to Yunnan, Xiaoming lives in Dali, Xiaoming and Xiaoqin are friends, these are the relationship between entities. Popular definition: knowledge graph is a relational network obtained by connecting all different kinds of information, so knowledge graph provides the ability to analyze problems from the perspective of "relationship".

The recommendation method based on graph knowledge of the present invention, after constructing a knowledge graph in the early stage and training an applicable model, can automatically combine user behavior data, output and store user behavior trajectories, extract key dimensions, and perform summary statistics for key dimensions. It can display the relationship between key entities in different dimensions, and can delineate the scope of different dimensions for detailed viewing and analysis.

Its nodes represent entities or concepts, and edges represent various semantic relationships between entities/concepts.

Please refer to FIG. 1 , which is a flowchart of a recommended method of the present invention. As shown in Figure 1, the recommended method based on graph knowledge of the present invention includes:

Step S1: Acquire the data provided by the user and organize the data. Wherein, the step S1 includes defining a label structure for the data. Specifically, the data includes: user data and material data, and the user data includes user behavior and user characteristics. The data is organized according to the user data and material data provided by the customer. If the provided data has clear and regular labels, it can be used directly. If the provided data is not clear, it needs to be reprocessed. First define the tag structure for the data: for example, determine how many tags to distinguish, and the course can have tags such as lecturer, price, classification, etc.; for example, what is the dictionary value of each tag and what is the range, such as the specific age of the user, which can be 10 years old as an interval range.

Step S2: Extracting graph elements from the sorted data and constructing knowledge graph data based on the association relationship between the graph elements.

Please refer to FIG. 2 , which is a flowchart of step S2 in FIG. 1 . As shown in Figure 2, the step S2 includes:

Step S21: Process the data through word segmentation and semantic processing;

Step S22: perform entity extraction, relation extraction and event extraction;

Step S23: Corresponding labels, associations and entities to form knowledge graph data.

Specifically, in this step, for information such as title, introduction, etc., word segmentation, semantic processing, etc. can be used. If basic information exists in the link, the basic information can be obtained through a crawler first, and then text processing is performed to associate tags, associations, etc. Relationships, etc. correspond to entities.

Relation extraction:

Definition: Automatically identify a certain semantic relationship between entities. According to the number of participating entities, it can be divided into binary relation extraction (two entities) and multivariate relation extraction (three or more entities).

By focusing on the semantic relationship between two entities, we can get (arg1, relation, arg2) triples, where arg1 and arg2 represent two entities, and relation represents the semantic relationship between entities. (For example, the semantic relationship between the words in the sentence can be obtained through the Hanlp analysis tool).

Relation extraction method:

Restricted domain relation extraction method:

Template-based relationship extraction method: Extract and discriminate entity relationships in text through manually edited or learned templates, which are limited by the quality and coverage of templates, and are not scalable. (Court documents made by yourself belong to template-based extraction)

Relation extraction method based on machine learning: regard relation extraction as a classification problem Among them, the relation extraction method based on machine learning can be divided into supervised and weakly supervised.

Supervised relation extraction methods:

Approaches based on feature engineering: need to explicitly convert relation instances into feature vectors acceptable to the classifier

Kernel function-based method: directly use the structure tree as the processing object, and instead of using the inner product of the feature vector when calculating the distance between the relationships, use the kernel function

Neural Network-Based Methods: Automatically Learn Efficient Feature Representations Directly from Input Text, End-to-End

Weakly supervised relation extraction methods: do not require manual labeling of large amounts of data.

Distance supervision: Automatic labeling of training samples with an open knowledge graph, which does not require manual labeling one by one, is a type of weakly supervised relationship extraction.

Open domain relation extraction method:

There is no need to predefine the relationship category, and some words in the entity pair context are used to describe the relationship between entities.

Event extraction and methods:

Definition: Extract events of interest to users from the text describing event information and present them in a structured form.

Steps: First identify the event and its type, secondly identify the elements (usually entities) involved in the event, and finally determine the role each element plays in the event.

Event extraction related concepts:

Event reference: a description of an objectively occurring specific event in the form of natural language, usually a sentence or a group of sentences

Event trigger word: refers to the word in an event reference that can best represent the occurrence of the event, and is an important feature that determines the category of the event, usually a verb or a noun

Event element: The participants in the event, mainly composed of entity, time and attribute value

Element role: What role does the event element play in the corresponding event

Event category: The event element and trigger word determine the category of the event (the category defines several subcategories)

Restricted domain event extraction: Before extraction, the type of target event and the specific structure of each type (which specific event elements are included) are pre-defined, and a certain amount of labeled data is usually given.

Qualified domain event extraction method:

Method based on pattern matching: The identification and extraction of certain types of events are carried out under the guidance of some patterns (steps: pattern acquisition, pattern matching)

Supervised Event Pattern Matching: Pattern acquisition is based entirely on human-annotated corpora

Weakly supervised event pattern matching: It is not necessary to fully label the corpus, and only need to manually pre-classify the corpus or formulate a small number of seed patterns

machine learning based methods

Supervised Event Extraction Methods: Modeling Event Extraction as a Multi-Classification Problem

Method based on feature engineering: It is necessary to explicitly convert event instances into feature vectors acceptable to the classifier, and the research focuses on how to extract discriminative features

Neural network-based method: automatically obtain features from text to complete event extraction, avoiding the error accumulation problem caused by using traditional natural language processing tools

Entity extraction:

Entity extraction or Named Entity Recognition (NER) plays an important role in information extraction, mainly extracting atomic information elements in text, such as person name, organization/institution name, geographic location, event/date, character value, amount value Wait. The entity extraction task has two keywords: find&classify, find named entities, and classify them.

Entity extraction methods are divided into three types: rule-based methods usually need to write templates for target entities, and then match them in the original corpus; statistical machine learning-based methods mainly train the original corpus through machine learning methods, and then Use the trained model to identify entities; open domain-oriented extraction will be oriented to massive Web corpus.

Step S3: Perform model training according to the knowledge graph data. Specifically, the user data and material data input model is used to train the model, and the recommendation result of the initial access is obtained.

Step S4: Obtain recommendation results through the trained model according to the real-time data. Specifically, in this step, after the trained model goes online, the recommendation result is output in combination with real-time user behavior data, that is, the recommendation result is output by continuously supplementing new users and real-time user data combined with real-time user behavior data. According to the user's new search for maternal and infant products within 1 hour, it recommends maternal and infant related products to users.

Step S5: output a single user association track in combination with the user's historical behavior and the recommendation result. Specifically, in this step, since the user's recommendation result is generally a black box for the customer, this part of the content can be converted into interpretable content, the historical behavior of the user referenced by the model is selected, and the different behaviors of the user can be recorded. Classification, for example, user A purchased a financial course (expand the price range, course duration, instructor information, etc.) of this financial course, and searched for an art course (also expanded the price range and course duration range of this art course) , instructor information, etc.), and clicked on an art course (identifying whether the click/other behavior is a recommended product for the user).

Analyze user behavior trajectories, study user preferences, and provide technical support for the active service model of "information seeking users". According to the analysis results of user behavior trajectories, obtain user preference information, recommend content that suits users' preferences, and realize the active service of "information seeking users". According to the evaluation mechanism of content recommendation, it is convenient to collect user evaluations of active services, and to facilitate the optimization of subsequent behavioral trajectory analysis models. Loop optimization of the trajectory analysis model, and feedback the optimization results to users to achieve closed-loop and sustainable accurate and active services.

The application scenarios are:

(1) Application content recommendation. During the application process, users find that there are too many dimension conditions, frequently used dimensions are not at the top, and the condition value must be reselected every time, which is inconvenient to operate. Through analysis, it is found that users often use only a few dimensions such as home city, time and brand. Solution: Recommend and display the dimensions that are frequently used by users, and hide the dimensions that are not frequently used. The recommended operation order can automatically perform the layout of the dimension display according to the order of dimensions frequently operated by the user, and automatically recommend and fill in the conditions that the user often selects or inputs. value.

(2) Pay attention to application recommendations. During the application process, the user finds that the system is too large and there are too many function nodes. Each time, he has to click many layers to find the function page he needs to use, and the applications concerned are scattered in different places, which is inconvenient to use. According to the user-focused application model, analyze the applications frequently accessed by users in their daily work, and find that there are too many functional nodes, and the tree nodes on the left side of the system are too large, which makes it very inconvenient for users to use, and the applications that users pay attention to are scattered in different places, which is inconvenient to use , and users often use only a few KPIs and ad-hoc query applications. Solution: Actively recommend the frequently used KPIs and ad hoc query applications to the user's personalized desk, or favorites, or the home page and other places where the user can directly use the application, so as to facilitate the user to use these applications.

(3) Auxiliary key application guarantee. Users found some key applications during the application process, and users reported that they often repeated the query many times, but could not find the data. Confirm key applications based on user-focused application models and system optimization models, and analyze the centralized access time period of key applications. Solution: The background UTAP adjusts the priority of tasks according to the user's access time requirements to ensure priority execution of key applications.

Step S6: extracting the data of each user for aggregation statistics and outputting the statistical results. Specifically, in this step, the user and behavioral characteristics of each user are extracted, and statistics are aggregated with different characteristics as the dimensions (for example, the summary data is viewed with the main dimensions such as region, category, price range, etc.), according to step S5 As a result, the user's tags are supplemented, such as the region information of a single user's behavior, demographic attribute information such as the user's gender, age, and job information such as the user's occupation, and then the data of the same user's behavior and the same tag are aggregated to obtain the user tag. The relationship between the recommended content and the recommended content, so as to obtain the recommendation reasons and recommendation effects of group users, which can be displayed on the interface in a visual way for customers to understand the behavioral preferences and recommendation effects of their own users, which can in turn guide the formulation of operational strategies And can assist targeted increase in promotional activities.

Please refer to FIG. 3 , which is a schematic diagram of the application of the recommended method of the present invention. As shown in Figure 3, the recommended method based on graph knowledge of the present invention is roughly the following process when applied:

1. Obtain the data provided by the customer for data processing;

2. Graph construction, including entity extraction, relation extraction and event extraction, etc.;

3. Use the constructed data for model training;

4. After the trained model goes online, it combines the real-time user behavior data to output the recommendation results;

5. Combine the user's historical behavior and recommendation results to output a single user's associated trajectory;

6. Extract the user and behavior characteristics of each user, and aggregate statistics with different characteristics as the dimensions (for example, view the aggregated data with the main dimensions such as region, category, and price range).

Please refer to FIG. 4 , which is a diagram showing an example of the recommendation method of the present invention. As shown in Fig. 4, the recommendation method of the present invention shows the correlation between previous behaviors and results in the form of association relationship, and it can be clearly seen that some behaviors and entities are closely related.

In another embodiment of the present invention, the labeling method of labels can also be used, and the possible association relationships can be viewed by combining multiple label dimensions.

Please refer to FIG. 5 , which is a schematic structural diagram of the recommendation system of the present invention. As shown in Figure 5, the recommendation system of the present invention includes:

Data processing unit 11: obtain the data provided by the user and organize the data, wherein, the data processing unit 11 defines a label structure for the data;

Construction unit 12: extracting map elements from the sorted data and constructing knowledge map data based on the relationship between the map elements;

Training unit 13: perform model training according to the knowledge graph data;

Result output unit 14: obtain recommendation results through the trained model according to real-time data;

Trajectory output unit 15: output a single user associated trajectory in combination with the user's historical behavior and recommendation results, wherein the trajectory output unit 15 selects the historical behavior of the user referenced by the model and records the classification of different user behaviors, according to the historical behavior, classification and the recommendation. The result outputs a single user associated trajectory;

Statistical result output unit 16: extracts the data of each user for aggregation statistics and outputs the statistical results, wherein the statistical result output unit 16 extracts the user behavior and user characteristics of each user, and uses different characteristics as dimensions to aggregate statistics and output the statistics result.

Wherein, the data includes: user data and material data, and the user data includes user behavior and user characteristics.

Further, after the construction unit 12 processes the data through word segmentation and semantic processing, the construction unit 12 performs entity extraction, relationship extraction and event extraction; then the construction unit 12 corresponds the tags, associations and entities to form knowledge graph data.

The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein the program executes the graph knowledge-based recommendation method described in any one of the above when the program runs .

The present invention also provides an electronic device comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to execute the graph-based mapping described in any one of the above by the computer program. Recommended methods of knowledge.

Specifically, the entity recommendation method of the embodiment of the present application described in conjunction with FIG. 1 may be implemented by an electronic device. FIG. 6 is a schematic diagram of the hardware structure of the electronic device of the present invention.

The electronic device may include a processor 81 and a memory 82 storing computer program instructions.

Specifically, the above-mentioned processor 81 may include a central processing unit (CPU), or a specific integrated circuit (Application Specific Integrated Circuit, ASIC for short), or may be configured as one or more integrated circuits implementing the embodiments of the present application.

Among others, memory 82 may include mass storage for data or instructions. By way of example and not limitation, the memory 82 may include a hard disk drive (Hard Disk Drive, abbreviated as HDD), a floppy disk drive, a solid state drive (referred to as SSD), flash memory, optical disk, magneto-optical disk, magnetic tape, or universal serial A bus (UniversalSerial Bus, referred to as USB) drive or a combination of two or more of these. Memory 82 may include removable or non-removable (or fixed) media, as appropriate. Where appropriate, memory 82 may be internal or external to the data processing device. In certain embodiments, memory 82 is a non-volatile (Non-Volatile) memory. In a specific embodiment, the memory 82 includes a read-only memory (Read-Only Memory, ROM for short) and a random access memory (Random Access Memory, RAM for short). In a suitable case, the ROM may be a mask-programmed ROM, a programmable ROM (Programmable Read-Only Memory, referred to as PROM), an erasable PROM (Erasable Programmable Read-Only Memory, referred to as EPROM), an electrically erasable Except PROM (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM for short), Electrically Rewritable ROM (Electrically Alterable Read-Only Memory, referred to as EAROM) or flash memory (FLASH) or a combination of two or more of these. In an appropriate case, the RAM may be Static Random-Access Memory (SRAM for short) or Dynamic Random Access Memory (DRAM for short), wherein the DRAM may be a fast page mode dynamic memory Random access memory (Fast PageMode Dynamic Random Access Memory, referred to as FPMDRAM), Extended Date Out Dynamic Random Access Memory (Extended Date Out Dynamic Random Access Memory, referred to as EDODRAM), Synchronous Dynamic Random Access Memory (Synchronous Dynamic Random-Access Memory) Access Memory, referred to as SDRAM) and so on.

Memory 82 may be used to store or cache various data files required for processing and/or communication use, as well as possibly computer program instructions executed by processor 81.

The processor 81 reads and executes the computer program instructions stored in the memory 82 to implement any one of the recommended methods in the above embodiments.

In some of these embodiments, the electronic device may also include a communication interface 83 and a bus 80 . Among them, as shown in FIG. 6 , the processor 81, the memory 82, and the communication interface 83 are connected through the bus 80 and complete the mutual communication.

The communication interface 83 is used to implement communication between modules, apparatuses, units and/or devices in the embodiments of the present application. The communication port 83 can also realize data communication with other components such as: external devices, image/data acquisition devices, databases, external storage, and image/data processing workstations.

The bus 80 includes hardware, software, or both, coupling the components of the electronic device to each other. The bus 80 includes, but is not limited to, at least one of the following: a data bus (Data Bus), an address bus (Address Bus), a control bus (Control Bus), an expansion bus (Expansion Bus), and a local bus (Local Bus). By way of example and not limitation, the bus 80 may include an Accelerated Graphics Port (AGP) or other graphics buses, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (Front Side Bus) , referred to as FSB), Hyper Transport (Hyper Transport, referred to as HT) interconnect, Industry Standard Architecture (Industry Standard Architecture, referred to as ISA) bus, wireless bandwidth (InfiniBand) interconnect, Low Pin Count (Low Pin Count, LPC bus, memory bus, MicroChannel Architecture (MCA) bus, Peripheral Component Interconnect (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology Attachment (Serial Advanced Technology Attachment, SATA for short) bus, Video Electronics Standards Association Local Bus (VLB for short) bus or other suitable bus or a combination of two or more of these. Where appropriate, bus 80 may include one or more buses. Although embodiments herein describe and illustrate a particular bus, this application contemplates any suitable bus or interconnect.

The electronic device can display the correlation between the previous behavior and the result in the form of an association relationship, so as to clearly see that some behaviors and entities are relatively closely related, and the close relationship can be bound in the subsequent target optimization scenarios. Relevant adjustments are made in a targeted manner, thereby realizing the recommendation method described in conjunction with FIG. 1 .

To sum up, the present invention can clearly see that some behaviors and entities are relatively closely related by showing the correlation between previous behaviors and results in the form of an association relationship, and this close relationship can be bound in subsequent goals. Targeted adjustments are made in optimized scenarios.

Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it is still possible to modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements to some of the technical features; and these Modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. A recommendation method based on atlas knowledge is characterized by comprising the following steps:

step S1: acquiring data provided by a user and performing data arrangement;

step S2: extracting map elements from the sorted data and constructing knowledge map data based on the incidence relation among the map elements;

step S3: performing model training according to the knowledge graph data;

step S4: and obtaining a recommendation result through the trained model according to the real-time data.

2. The graph-knowledge-based recommendation method of claim 1, further comprising:

step S5: and outputting a single user association track by combining the historical behaviors of the users and the recommendation result.

3. The graph-knowledge based recommendation method of claim 2, further comprising:

step S6: and extracting data of each user for aggregation statistics and outputting a statistical result.

4. The graph-knowledge based recommendation method of claim 3, wherein the data comprises: user data and material data, the user data including user behavior and user characteristics.

5. The graph knowledge-based recommendation method according to any one of claims 1-4, wherein said step S1 includes defining a tag structure for said data.

6. The graph knowledge-based recommendation method according to claim 5, wherein the step S2 comprises:

step S21: processing the data through word segmentation and semantic processing;

step S22: performing entity extraction, relationship extraction and event extraction;

step S23: and the labels, the association relation and the entities are corresponded to form knowledge graph data.

7. The method for recommending based on atlas knowledge according to claim 2, wherein the step S5 includes selecting the historical behaviors of the user referenced by the model and recording the classification of different behaviors of the user at the same time, and outputting a single user association track according to the historical behaviors, the classification and the recommendation result.

8. The graph knowledge-based recommendation method according to claim 3, wherein the step S6 comprises extracting user behaviors and user characteristics of each user, performing aggregation statistics with different characteristics as dimensions, and outputting the statistical results.

9. A graph knowledge-based recommendation system, comprising:

a data processing unit: acquiring data provided by a user and performing data arrangement;

a construction unit: extracting map elements from the sorted data and constructing knowledge map data based on the incidence relation among the map elements;

a training unit: performing model training according to the knowledge graph data;

a result output unit: obtaining a recommendation result through the trained model according to the real-time data;

a trajectory output unit: outputting a single user association track by combining the historical behaviors of the users and the recommendation result;

a statistical result output unit: and extracting data of each user for aggregation statistics and outputting a statistical result.

10. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the graph knowledge-based recommendation method according to any one of claims 1 to 8 by the computer program.

Images