WO2022032685A1

WO2022032685A1 - Method and device for constructing multi-level knowledge graph

Info

Publication number: WO2022032685A1
Application number: PCT/CN2020/109365
Authority: WO
Inventors: Armin Roux; Bin Zhang; Shunjie Fan; Ming JIE; Zhongyang SUN
Original assignee: Siemens Aktiengesellschaft; Siemens Ltd., China
Priority date: 2020-08-14
Filing date: 2020-08-14
Publication date: 2022-02-17
Also published as: CN116097253A

Abstract

The present disclosure relates to the field of artificial intelligence technology, and in particular to a method and device for constructing multi-level knowledge graph. The method comprising: constructing (102) a first-level knowledge graph containing parent nodes and association relationships between parent nodes; constructing (104) a second-level knowledge graph of a parent node, where the second-level knowledge graph containing child nodes belonging to the parent node and association relationships between the child nodes; nesting (106) the second-level knowledge graph into the parent node of the first-level knowledge graph.

Description

Method and device for constructing multi-level knowledge graph

FIELD

The present invention relates to the field of artificial intelligence technology, in particular to a method and device for constructing multi-level knowledge graph.

BACKGROUND

In the field of artificial intelligence, knowledge graphs are often used to build cognition and easily understand the world. Intelligent and efficient knowledge graph is very necessary for industrial digitization. The basic function of knowledge graph is to realize an organic integration of data, but current technology is mainly limited to a single type of structured data, and there is no intelligent knowledge graph that can integrate multiple data types.

Moreover, the present knowledge graph has only one level of association network, and it is difficult to provide enough associated information.

In addition, the present knowledge graph that assists in providing suggestions usually only works at the back end, which makes it impossible for users to understand the connection between different suggestions and the reason behind the recommendations when receiving suggestions.

SUMMARY

The embodiment of the present invention proposes a method and device for constructing multi-level knowledge graph.

In a first aspect, a method for constructing a multi-level knowledge graph is provided. The method comprising:

constructing a first-level knowledge graph containing parent nodes and association relationships between parent nodes;

constructing a second-level knowledge graph of a parent node, where the second-level knowledge graph containing child nodes belonging to the parent node and association relationships between the child nodes;

nesting the second-level knowledge graph into the parent node of the first-level knowledge graph.

Therefore, a multi-layer knowledge graph is provided to enrich the associated knowledge.

Preferably, comprising:

constructing a third-level knowledge graph of a child node, wherein the third-level knowledge graph containing grandchildren belonging to the same child node and association relationship between the grandchildren;

nesting the third-level knowledge graph into the child node of the second-level knowledge graph.

Therefore, it further provides a deeper knowledge graph and further enriches the associated knowledge.

Preferably, wherein the first-level knowledge graph further comprising: the tributes of the parent nodes and association relationships between the tributes of the parent nodes and the parent nodes; the second-level knowledge graph further comprising: the tributes of the child nodes and association relationships between the tributes of the child nodes and the child nodes.

Preferably, wherein the data sources used to construct the first-level knowledge graph, the second-level knowledge graph or the third-level knowledge graph include one of the following:

structured data; unstructured data; semi-structured data.

Preferably, wherein the nested knowledge graph further comprising: association relationships between child nodes belonging to different parent nodes.

Preferably, comprising:

displaying recommendation information based on the nested knowledge graph, and displaying the first-level knowledge graph when the recommendation information is only related to parent nodes; or

displaying recommendation information based on the nested knowledge graph, and displaying the nested knowledge graph of a single parent node when the recommendation information is related to the single parent node and the child nodes of the single parent node; or

displaying recommendation information based on the nested knowledge graph and displaying respective nested knowledge graphs of multiple parent nodes when the recommendation information is related to multiple parent nodes and child nodes of the parent nodes.

In a second aspect, a device (600) for constructing a multi-level knowledge graph is provided. The device comprising:

a first constructing module, configured to construct a first-level knowledge graph containing parent nodes and association relationships between parent nodes;

a second constructing module, configured to construct a second-level knowledge graph of a parent node, where the second-level knowledge graph containing child nodes belonging to the parent node and association relationships between the child nodes; and

a nesting module, configured to nest the second-level knowledge graph into the parent node of the first-level knowledge graph.

Preferably, comprising:

a third constructing module, configured to construct a third-level knowledge graph of a child node, wherein the third-level knowledge graph containing grandchildren belonging to the same child node and association relationship between the grandchildren;

wherein the nesting module, configured to nest the third-level knowledge graph into the child node of the second-level knowledge graph.

structured data; unstructured data; semi-structured data.

Preferably, comprising:

a displaying module, configured to:

display recommendation information based on the nested knowledge graph, and display the first-level knowledge graph when the recommendation information is only related to parent nodes; or

display recommendation information based on the nested knowledge graph, and display the nested knowledge graph of a single parent node when the recommendation information is related to the single parent node and the child nodes of the single parent node; or

display recommendation information based on the nested knowledge graph and display respective nested knowledge graphs of multiple parent nodes when the recommendation information is related to multiple parent nodes and child nodes of a parent node.

A device for constructing a multi-level knowledge graph is provided. The device comprising a processor and a memory, wherein an application program executable by the processor is stored in the memory for causing the processor to execute a method for constructing a multi-level knowledge graph according to any one of above.

A computer-readable medium comprising computer-readable instructions stored thereon is provided, wherein the computer-readable instructions for executing a method for constructing a multi-level knowledge graph according to any one of above.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to make technical solutions of examples of the present disclosure clearer, accompanying drawings to be used in description of the examples will be simply introduced hereinafter. Obviously, the accompanying drawings to be described hereinafter are only some examples of the present disclosure. Those skilled in the art may obtain other drawings according to these accompanying drawings without creative labor.

Fig. 1 is a flowchart of a method for constructing a multi-level knowledge graph in an embodiment of the present invention.

FIG. 2 is an exemplary schematic diagram of constructing a single-layer knowledge graph according to an embodiment of the present invention.

FIG. 3 is an exemplary schematic diagram of the first-level knowledge graph of the embodiment of the present invention.

FIG. 4 is an exemplary schematic diagram of the second-level knowledge graph of the embodiment of the present invention.

FIG. 5 is an exemplary schematic diagram of nesting the second-level knowledge graph into a parent node in an embodiment of the present invention.

FIG. 6 is an exemplary schematic diagram showing the association relationship between parent nodes in an embodiment of the present invention.

FIG. 7 is a schematic diagram of a first visualization interface displaying recommendation information in an embodiment of the present invention.

FIG. 8 is a schematic diagram of a second visualization interface displaying recommendation information in an embodiment of the present invention.

FIG. 9 is a schematic diagram of a third visualization interface displaying recommendation information in an embodiment of the present invention.

Fig. 10 is an exemplary structure diagram of a device for constructing a multi-level knowledge graph according to an embodiment of the present invention.

FIG. 11 is an exemplary structure diagram of a device for constructing a multi-level knowledge graph according to an embodiment of the present invention.

List of reference numbers:

DETAILED DESCRIPTION

In order to make the technical solutions and advantages of the present invention more comprehensible, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the scope of the invention.

For the sake of brevity and clarity of the description, the aspects of the present invention are set forth below by describing several representative embodiments. Numerous details in the embodiments are only configured to assist in understanding the aspects of the present invention. However, it is obvious that the technical solution of the present invention can be implemented without being limited to these details. In order to avoid unnecessarily obscuring aspects of the present invention, some embodiments are not described in detail, but only the framework is given. Hereinafter, "including" means "including but not limited to" , and "according to" means "at least according to ..., but not limited to only based on" . Due to the language habit of Chinese, the number of one component is not specifically indicated below, which means that the component may be one or more, or may be understood as a.

As shown in Figure 1, the method includes:

Step 102: constructing a first-level knowledge graph containing parent nodes and association relationships between parent nodes.

Here, the process of constructing the first-level knowledge graph can include manual construction, automatic construction, and semi-automatic construction.

In one embodiment, the process of automatically constructing the first-level knowledge graph includes:

First, obtain a data source, where the data source includes multiple entities. For example, the data source can be text data such as news, posts, popular articles, etc. The text data can be in the form of a data table or in other forms. The data source can be called a collection of corpuses. The data source can be obtained by crawling online news, announcements, legal documents, industrial and commercial websites, corporate official websites, personal homepages, etc. by crawling. The data source can also be any device that can be used to collect and send data, such as a terminal device, which can be a smart phone, a tablet computer, a laptop computer, a desktop computer, or a crawler server.

Then, semantic analysis and cluster analysis are performed on the data source, and the entity set and attribute set are extracted from the data source. Wherein, the attribute set includes the entity attributes of each entity in the entity set. For example, when creating an enterprise knowledge graph, entities can refer to employee names and enterprise names, and entity attributes refer to enterprise attributes and employee attributes. Among them, the entity attributes of employees can be employee position, employee gender, employee education, award information, employee level, employee resume, and so on. The entity attributes of a company can be information such as announcements, news, legal documents, intellectual property rights, products, qualifications, official websites, recruitment, administrative penalties, research teams and events, stock codes, shareholder information, investments, and executives. Semantic analysis refers to semantic checking and processing according to the grammatical category recognized by the grammar analyzer to obtain the substantial meaning of the text. Cluster analysis refers to the analysis process of grouping a collection of physical or abstract objects into multiple classes composed of similar objects.

Then, the association relationship between each entity and attribute in the entity set is obtained. Create and construct a first-level knowledge graph according to the entity set, attribute set, and the association relationship between entities and attributes. Among them, the knowledge graph may specifically include entities, entity attributes, association relationships between entities and attributes, and association relationships between entities.

In one embodiment, the process of manually constructing the first-level knowledge graph includes: defining specifications based on industry knowledge to construct the first-level knowledge graph. For example: define several types of entities: people, tasks, skills, machines, parts, etc.

In one embodiment, the process of semi-automatically constructing the first-level knowledge graph includes: extracting entity sets and attribute sets based on description information. Wherein, the attribute set includes entity attributes of each entity in the entity set. For example, assuming there are multiple video files, the description information or comment information of each video file is automatically obtained, and semantic analysis and clustering analysis are performed on the description information or comment information to extract the entity set and attribute set, and construct the parent node (that is, Each video file) and the first level of knowledge graph of the first association between parent nodes.

Step 104: constructing a second-level knowledge graph of a parent node, where the second-level knowledge graph containing child node s (sub-nodes) belonging to the parent node and association relationships between the child nodes.

Preferably, the process of constructing the second-level knowledge graph of each parent node is an automatic construction process.

Specifically, the process of constructing the second-level knowledge graph of each parent node is an automatic construction process including: obtaining data sources related to the parent node. Then, semantic analysis and cluster analysis are performed on the data source, and the entity set and attribute set are extracted from the data source. Among them, the attribute set includes the entity attributes of each entity in the entity set. Then, the association relationship between each entity and attribute in the entity set is obtained. Create and construct a second-level knowledge graph according to the entity set, attribute set, and the association relationship between entities and attributes.

For example, when a certain parent node of the first-level knowledge graph is a video file and data sources related to the video file can be obtained, such as screenshots and audio extraction in the video file. Then, perform image recognition processing for the screenshot, input the image recognition result (text description of the image) into the NLP-based semantic analysis system; perform voice recognition processing on the extracted audio, and input the voice recognition result (text content) into the NLP-based semantic analysis system Semantic analysis system. The NLP-based semantic analysis system performs NLP processing on these multi-channel text inputs to extract ontology data. Then, the knowledge graph of the video file as the parent node can be constructed based on the ontology data, that is, the second-level knowledge graph.

Step 106: nesting the second-level knowledge graph into the parent node of the first-level knowledge graph.

Here, the second-level knowledge graph is filled into the at least one parent node of the first-level knowledge graph, thereby forming a nested knowledge graph.

Preferably, the sub-nodes can be further split to form a knowledge graph of more levels.

In one embodiment, the method further includes: constructing a third-level knowledge graph of at least one child node, wherein the third-level knowledge graph includes grandchildren belonging to the same child node and a third association relationship between the grandchildren ; Nesting the third-level knowledge graph into the at least one child node of the second-level knowledge graph.

In one embodiment, the first-level knowledge graph further includes: the attributes of the parent node and the fourth association relationship between the attributes of the parent node and the parent node; the second-level knowledge graph further includes: the attributes of the child nodes And the fifth association relationship between the child node and the attributes of the child node.

In one embodiment, the data source used to construct the first-level knowledge graph, the second-level knowledge graph, or the third-level knowledge graph includes at least one of the following: structured data; unstructured data; semi-structured data.

In an embodiment, the nested knowledge graph further includes: an association relationship between child nodes belonging to different parent nodes.

In one embodiment, the method further includes: displaying recommendation information based on the nested knowledge graph, wherein when the recommendation information is only related to the parent node, displaying the first-level knowledge graph; or, based on the nested knowledge graph. The nested knowledge graph displays recommendation information, where when the recommendation information is related to a single parent node and the child nodes of the single parent node, the nested knowledge graph of the single parent node is displayed; or, based on the nested post The knowledge graph of shows recommendation information, where when the recommendation information is related to multiple parent nodes and child nodes of at least one parent node, the respective nested knowledge graphs of multiple parent nodes are displayed. Here, the recommendation information can be presented in multiple ways such as video, audio, photos or text.

In one embodiment, structured user data can be obtained from structured databases such as a personnel resource database, a work log database, etc., as a data source for constructing a knowledge graph. Optionally, unstructured user data can also be obtained from unstructured data sources, etc., as a data source for constructing a knowledge graph.

Examples of the present invention are described below with examples. Assume that video file 1, video file 2, video file 3, and video file 4 exist. First, the description information of each video file is proposed from the description information database of these video files, and the first-level knowledge graph is proposed based on the description information. The first-level knowledge graph contains the main node (each video file) and the relationship between each video file. Moreover, for each video file, extract the audio separately, perform voice recognition on the audio to obtain the text result, and then perform NLP processing on the text result to obtain the child nodes belonging to each video file. For example, the child nodes include actors or lines. and many more. Establish a second-level knowledge graph of each video file, the second-level knowledge graph contains sub-nodes and the relationship between sub-nodes. Then, fill the second-level knowledge graph of each video file into the respective main nodes of the first-level knowledge graph.

Fig. 2 is an exemplary schematic diagram of constructing a knowledge graph in an embodiment of the present invention.

As shown in FIG. 2, the data source 20 includes multiple types, specifically including document 21, image 22, audio 23 and video 24. Among them, texts contained in the document 21 can be input to NLP-based semantic analysis system 30. The image recognition 25 performs image recognition processing on the picture 22, and inputs the image recognition result (text description of the image) into NLP-based semantic analysis system 30. The voice recognition 27 performs voice recognition processing on the audio 23, and inputs the voice recognition result into NLP-based semantic analysis system 30. Audio is first extracted from video 24 based on the audio extraction 28, then speech recognition 29 is performed on the extracted audio, and result of the speech recognition is input to NLP-based semantic analysis system 30.

It can be seen that the NLP-based semantic analysis system 30 has multiple text input sources. The NLP-based semantic analysis system 30 executes NLP processing to extract ontology data 60. Then, based on the ontology data 60, a knowledge graph including user entities, skill entities, and operation target entities can be created. For example, tools such as Neo4j or MongoDB can be used to create knowledge graphs. Among them, the user entity may contain a triple represented as <user ID, user attribute, user attribute value>; the skill entity may contain a triple represented as <skill ID, skill attribute, skill attribute value>; the operation target entity may contain triples represented as < operation target identifier, operation target attribute, operation target attribute value>. Preferably, the knowledge graph is in a dynamic update state.

In one embodiment, wherein determining recommendation information corresponding to the user identifier based on the score value and a knowledge graph related to the user comprises: determining a skill set corresponding to the level range to which the score value belongs; determining a skill corresponding to the user identifier stored in the knowledge graph; removing the skill corresponding to the user identifier from the skill set; and determining the recommendation information corresponding to the user identifier based on the remaining skills in the skill set.

In one embodiment, wherein determining recommendation information corresponding to the user identifier based on the score value and a knowledge graph related to the user comprises: determining a similar user with a score value similar to the score value; determining a skill corresponding to the user identifier of the similar user stored in the knowledge graph; and determining the recommendation information corresponding to the user identifier based on the skill corresponding to the user identifier of the similar user.

In one embodiment, wherein determining recommendation information corresponding to the user identifier based on the score value and a knowledge graph related to the user comprises: determining a similar user who is similar to the user corresponding to the user identifier based on the knowledge graph; determining a skill corresponding to the user identifier of the similar user stored in the knowledge graph; determining the recommendation information corresponding to the user identifier based on the skill corresponding to the user identifier of the similar user.

In one embodiment, wherein determining recommendation information corresponding to the user identifier based on the score value and a knowledge graph related to the user comprises: determining a first set of similar users based on the knowledge graph, wherein the first set of similar users includes similar users of the user corresponding to the user identifier; determining a second set of similar users based on a score value comparison process, wherein the second set of similar users includes similar users of the user corresponding to the user identifier; determining the intersection of the first set of similar users and the second set of similar users; and determining the recommendation information corresponding to the user identifier based on a skill of the users in the intersection stored in the knowledge graph and a skill of the user corresponding to the user identifier stored in the knowledge graph.

FIG. 3 is an exemplary schematic diagram of the first-level knowledge graph of the embodiment of the present invention. In Figure 3, the first-level knowledge graph is shown, including multiple parent nodes, namely parent node A, parent node B, parent node C, parent node D, parent node E, parent node F, parent node G, parent Node H, parent node I, parent node J, and parent node K. In the first-level knowledge graph, parent node A, parent node B, parent node C, parent node D, parent node E, parent node F, parent node G, parent node H, parent node I, parent node J are also displayed Interrelationship with parent node K. Specifically, these parent nodes may all be video files, and the description of the video file is used to create the first-level knowledge graph.

FIG. 4 is an exemplary schematic diagram of the second-level knowledge graph of the embodiment of the present invention. In Figure 4, the knowledge graph of parent node A in Figure 3 is shown. It can be seen that parent node A includes child node a, child node b, child node c, child node d, and child node e. In Figure 4, the correlation between child node a, child node b, child node c, child node d, and child node e is also shown. Specifically, for the video file of the parent node A, screenshots and audio can be extracted separately. Then, perform image recognition processing for the screenshot, input the image recognition result (text description of the image) into the NLP-based semantic analysis system; perform voice recognition processing on the extracted audio, and input the voice recognition result (text content) into the NLP-based semantic analysis system Semantic analysis system. The NLP-based semantic analysis system performs NLP processing on these multi-channel text inputs to extract ontology data. Then, the knowledge graph of the video file as the parent node can be constructed based on the ontology data, that is, the second-level knowledge graph of the parent node A.

Similarly, for other parent nodes (video files) in the first-level knowledge graph, you can also create their own second knowledge graphs.

FIG. 5 is an exemplary schematic diagram of nesting the second-level knowledge graph into a parent node in an embodiment of the present invention. It can be seen that in Fig. 5, the second-level knowledge graph shown in Fig. 4 is filled into the parent node A of Fig. 3. Among them, the illustration of the rest of the first-level knowledge graph of FIG. 3 except for the parent node A is omitted.

FIG. 6 is an exemplary schematic diagram showing the association relationship between parent nodes in an embodiment of the present invention. It can be seen from Fig. 6 that the association relationship between parent nodes can be displayed through the association relationship of child nodes, and part of the association relationship can also be highlighted.

FIG. 7 is a schematic diagram of a first visualization interface displaying recommendation information in an embodiment of the present invention. In FIG. 7, recommendation information 71 is displayed based on the nested knowledge graph, wherein when the recommendation information is only related to the parent node, the first-level knowledge graph 81 is displayed.

FIG. 8 is a schematic diagram of a second visualization interface displaying recommendation information in an embodiment of the present invention. In FIG. 8, recommendation information 81 is displayed based on the nested knowledge graph. It can be seen that the recommendation information 81 is related to a single parent node and the child nodes of the single parent node, so the nested knowledge graph 82 of the single parent node is displayed.

FIG. 9 is a schematic diagram of a third visualization interface displaying recommendation information in an embodiment of the present invention. In FIG. 9, recommendation information 91 is displayed based on the nested knowledge graph. It can be seen that the recommendation information 91 is related to multiple parent nodes and child nodes of at least one parent node, showing the respective nested knowledge graphs of multiple parent nodes 92.

Based on the above description, the embodiment of the present invention also proposes a device for constructing a multi-level knowledge graph.

Fig. 10 is a block diagram of a device for constructing a multi-level knowledge graph according to an embodiment of the present invention.

As shown in FIG. 10, the device 600 for constructing a multi-level knowledge graph includes: a first constructing module 601, configured to construct a first-level knowledge graph containing parent nodes and association relationships between parent nodes; a second constructing module 602, configured to construct a second-level knowledge graph of a parent node, where the second-level knowledge graph containing child nodes belonging to the parent node and association relationships between the child nodes; and a nesting module 603, configured to nest the second-level knowledge graph into the parent node of the first-level knowledge graph.

In one embodiment, the device 600 comprising: a third constructing module 604, configured to construct a third-level knowledge graph of a child node, wherein the third-level knowledge graph containing grandchildren belonging to the same child node and association relationship between the grandchildren; wherein the nesting module 603, configured to nest the third-level knowledge graph into the child node of the second-level knowledge graph.

In one embodiment, wherein the first-level knowledge graph further comprising: the tributes of the parent nodes and association relationships between the tributes of the parent nodes and the parent nodes; the second-level knowledge graph further comprising: the tributes of the child nodes and association relationships between the tributes of the child nodes and the child nodes.

In one embodiment, wherein the data sources used to construct the first-level knowledge graph, the second-level knowledge graph or the third-level knowledge graph include one of the following: structured data; unstructured data; semi-structured data.

In one embodiment, wherein the nested knowledge graph further comprising: association relationships between child nodes belonging to different parent nodes.

In one embodiment, the device 600 comprising a displaying module 605, configured to:

display recommendation information based on the nested knowledge graph and display respective nested 11knowledge graphs of multiple parent nodes when the recommendation information is related to multiple parent nodes and child nodes of a parent node.

FIG. 11 is an exemplary structure diagram of a device for constructing a multi-level knowledge graph according to an embodiment of the present invention

As shown in FIG. 11 the device 700 includes a processor 701, a memory 702, and a computer program stored on the memory 702 and running on the processor 701. The computer program is executed by the processor 701 to perform any above method for constructing a multi-level knowledge graph.

Among them, the memory 702 may be specifically implemented as various storage media such as an electrically erasable programmable read-only memory (EEPROM) , a flash memory (Flash memory) , and a programmable program read-only memory (PROM) . The processor 701 may be implemented to include one or more central processing units or one or more field programmable gate arrays, where the field programmable gate array integrates one or more central processing unit cores. Specifically, the central processing unit or central processing unit core can be implemented as a CPU, MCU, DSP, or the like.

It can be seen that the present disclosure can integrate different types of knowledge into one knowledge graph and push them at the same time. The user can know the connection of each part through the knowledge graph while receiving the push content. Users can view the various components of the push item and visually understand the relationship between different levels through the deep knowledge graph. Users can interactively adjust the knowledge graph to make the recommendation system more personalized.

It should be noted that not all the steps and modules in the foregoing processes and the various structural diagrams are necessary, and some steps or modules may be omitted according to actual needs. The order of execution of each step is not fixed and can be adjusted as needed. The division of each module is only for the convenience of description of the functional division. In actual implementation, one module can be implemented by multiple modules, and the functions of multiple modules can also be implemented by the same module. These modules can be located in the same device. It can also be located in different devices.

The hardware modules in the various embodiments may be implemented mechanically or electronically. For example, a hardware module can include specially designed permanent circuits or logic devices (such as dedicated processors such as FPGAs or ASICs) for performing specific operations. The hardware modules may also include programmable logic devices or circuits (such as including general purpose processors or other programmable processors) that are temporarily configured by software for performing particular operations. The hardware module can be implemented by mechanical means, by using a dedicated permanent circuit, or by using a temporarily configured circuit (such as software configuration) , which can be determined based on cost and time considerations.

The present invention also provides a machine readable storage medium storing instructions for causing a machine to perform a method as described herein. In particular, a system or apparatus equipped with a storage medium on which software program code implementing the functions of any of the above-described embodiments is stored, and a computer (or CPU or MPU) of the system or apparatus may be stored Reading and executing the program code stored in the storage medium. In addition, some or all of the actual operations may be performed by an operating system or the like operating on a computer based on instructions of the program code. It is also possible to write the program code read out from the storage medium into a memory set in an expansion board inserted into the computer or into a memory set in an extension unit connected to the computer, and then install the program based on the instruction of the program code. The expansion board or the CPU or the like on the expansion unit performs part and all of the actual operations to implement the functions of any of the above embodiments.

Embodiments of storage medium for providing program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW) , Tape, non-volatile memory card and ROM. Alternatively, the program code can be downloaded from a server computer or cloud by a communication network.

The above is only preferred embodiments of the present invention and is not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and scopes of the present invention are intended to be included within the scope of the present invention.

The present invention has been shown and described in detail with reference to the accompanying drawings and the preferred embodiments thereof, but the invention is not limited to these disclosed embodiments, and those skilled in thert may know that the various embodiments described above may be combined. The code review means in the present invention obtains more embodiments of the present invention, and these embodiments are also within the scope of the present invention.

Claims

A method (100) for constructing a multi-level knowledge graph, comprising:

constructing (102) a first-level knowledge graph containing parent nodes and association relationships between parent nodes;

constructing (104) a second-level knowledge graph of a parent node, where the second-level knowledge graph containing child nodes belonging to the parent node and association relationships between the child nodes;

nesting (106) the second-level knowledge graph into the parent node of the first-level knowledge graph.
The method (100) according to claim 1, comprising:

constructing a third-level knowledge graph of a child node, wherein the third-level knowledge graph containing grandchildren belonging to the same child node and association relationship between the grandchildren;

nesting the third-level knowledge graph into the child node of the second-level knowledge graph.
The method (100) according to claim 1, wherein the first-level knowledge graph further comprising: the tributes of the parent nodes and association relationships between the tributes of the parent nodes and the parent nodes; the second-level knowledge graph further comprising: the tributes of the child nodes and association relationships between the tributes of the child nodes and the child nodes.
The method (100) according to claim 2, wherein the data sources used to construct the first-level knowledge graph, the second-level knowledge graph or the third-level knowledge graph include one of the following:

structured data; unstructured data; semi-structured data.
The method (100) according to claim 1, wherein the nested knowledge graph further comprising: association relationships between child nodes belonging to different parent nodes.
The method (100) according to claim 5, comprising:

displaying recommendation information based on the nested knowledge graph, and displaying the first-level knowledge graph when the recommendation information is only related to parent nodes; or

displaying recommendation information based on the nested knowledge graph, and displaying the nested knowledge graph of a single parent node when the recommendation information is related to the single parent node and the child nodes of the single parent node; or

displaying recommendation information based on the nested knowledge graph and displaying respective nested knowledge graphs of multiple parent nodes when the recommendation information is related to multiple parent nodes and child nodes of the parent nodes.
A device (600) for constructing a multi-level knowledge graph, comprising:

a first constructing module (601) , configured to construct a first-level knowledge graph containing parent nodes and association relationships between parent nodes;

a second constructing module (602) , configured to construct a second-level knowledge graph of a parent node, where the second-level knowledge graph containing child nodes belonging to the parent node and association relationships between the child nodes; and

a nesting module (603) , configured to nest the second-level knowledge graph into the parent node of the first-level knowledge graph.
The device (600) according to claim 7, comprising:

a third constructing module (604) , configured to construct a third-level knowledge graph of a child node, wherein the third-level knowledge graph containing grandchildren belonging to the same child node and association relationship between the grandchildren;

wherein the nesting module (603) , configured to nest the third-level knowledge graph into the child node of the second-level knowledge graph.
The device (600) according to claim 7, wherein the first-level knowledge graph further comprising: the tributes of the parent nodes and association relationships between the tributes of the parent nodes and the parent nodes; the second-level knowledge graph further comprising: the tributes of the child nodes and association relationships between the tributes of the child nodes and the child nodes.
The device (600) according to claim 8, wherein the data sources used to construct the first-level knowledge graph, the second-level knowledge graph or the third-level knowledge graph include one of the following:

structured data; unstructured data; semi-structured data.
The device (600) according to claim 7, wherein the nested knowledge graph further comprising: association relationships between child nodes belonging to different parent nodes.
The device (600) according to claim 11, comprising:

a displaying module (605) , configured to:

display recommendation information based on the nested knowledge graph, and display the first-level knowledge graph when the recommendation information is only related to parent nodes; or

display recommendation information based on the nested knowledge graph, and display the nested knowledge graph of a single parent node when the recommendation information is related to the single parent node and the child nodes of the single parent node; or

display recommendation information based on the nested knowledge graph and display respective nested knowledge graphs of multiple parent nodes when the recommendation information is related to multiple parent nodes and child nodes of a parent node.
A device (700) for constructing a multi-level knowledge graph, comprising a processor (701) and a memory (702) , wherein an application program executable by the processor (701) is stored in the memory (702) for causing the processor (701) to execute a method (100) for constructing a multi-level knowledge graph according to any one of claims 1 to 6.
A computer-readable medium comprising computer-readable instructions stored thereon, wherein the computer-readable instructions for executing a method (100) for constructing a multi-level knowledge graph according to any one of claims 1 to 6.