CN110750649A

CN110750649A - Knowledge graph construction and intelligent response method, device, equipment and storage medium

Info

Publication number: CN110750649A
Application number: CN201810739557.5A
Authority: CN
Inventors: 艾华东; 朱石争
Original assignee: ZTE Corp
Current assignee: ZTE Corp
Priority date: 2018-07-06
Filing date: 2018-07-06
Publication date: 2020-02-04
Also published as: WO2020007224A1

Abstract

The invention provides a knowledge graph construction and intelligent response method, a knowledge graph construction and intelligent response device, knowledge graph construction equipment and a storage medium. The whole establishing process is not limited by a concept layer any more, the degree of freedom is high, and the whole knowledge graph is convenient to construct and adjust. For example, when new attribute information is added to a created entity, the new attribute information can be directly added without being limited by a concept layer. In addition, in the embodiment of the disclosure, the entity is directly created, so that there is no requirement for refining the upper-layer concepts and attributes, which reduces the knowledge requirement for the knowledge graph builder, and even if the builder does not have strong abstraction ability, a better knowledge graph can be constructed, which has better universality.

Description

Knowledge graph construction and intelligent response method, device, equipment and storage medium

Technical Field

The present disclosure relates to, but not limited to, the field of data processing technologies, and in particular, to, but not limited to, a method, an apparatus, a device, and a storage medium for knowledge graph construction and intelligent response.

Background

Knowledge Graph (knowledgegraph), also known as scientific Knowledge Graph or semantic web, is intended to describe various entities or concepts and their relationships that exist in the real world, and constitutes a huge semantic network Graph, with nodes representing entities or concepts and edges consisting of attributes or relationships. With the development of the internet, the content of the network data presents an explosive growth situation. Due to the characteristics of large scale, heterogeneous multiple and loose organization structure of internet content, the method provides challenges for people to effectively acquire information and knowledge. The primary intention of knowledge graph creation is to improve the ability of search engines, improve the search quality and search experience for users. With the technical development and application of artificial intelligence, the knowledge graph is one of key technologies, can provide semantic understanding and reasoning ability, and is widely applied to the fields of intelligent search, intelligent question answering, personalized recommendation, content distribution and the like.

The current knowledge graph generally comprises elements such as concepts (also called classes or ontologies), entities (also called individuals), attributes, relations and the like. Concepts refer primarily to collections, categories, types of objects, categories of things, etc., such as people, geographies, etc. An entity refers to something that is distinguishable and exists independently; such as a person, an item, etc. All things in the world are composed of specific things, which are entities. The entity is the most basic element in the knowledge graph, and different relationships exist among different entities. Attributes are unique information pertaining to concepts and entities. Such as a person having attributes of gender, height, weight, age, etc. Relationships are relationships that describe relationships between entities or between concepts. For example, there are relationships between people, such as friends, couples, and colleagues, and there is an employee relationship between people and companies. There are some relations, such as superior, inferior; some relationships have no direction, such as colleagues, friends, etc.; two relationships with directions may be opposite to each other, such as parent and child.

At present, when a knowledge graph is constructed, a concept layer is constructed first, and then a physical layer is constructed according to the structure of the concept layer after the concept layer is constructed. For example, referring to FIG. 1, FIG. 1 is a knowledge graph constructed in the present manner. The construction process comprises the following steps:

s101: creating the concept "team";

s102: creating the concept "player";

s103: creating a 'player' relationship between two concepts of 'team' and 'player';

s104: adding the 'native' attribute of the 'player' concept;

each concept may define a variety of different attributes such as the native whereabouts, weight, height, number, location on the court, etc. of the player. The specific entity of the concept inherits the attribute of the concept and assigns a specific attribute value to the attribute.

S105: creating an entity "Hengda" representing a port team on the sea;

the method is similarly created for a plurality of entities of 'Guoan' and 'Shangang'.

S106: adding inheritance relation connecting lines between concepts and entities;

the representation of an entity "constant big" here is an example of the concept "team". There may be multiple instances of a concept, such as the concept "team" with "hong kong", "national security", "constant" entities in fig. 1, and the concept "player" with "zhangzheng", "Gao forest", "zheng zhi", "high mingjie".

S107: creating an entity "Gao forest" inherited from the concept "player";

it should be understood that similar entities such as "Zhang Zheng", "Gaogejing", "Zheng Zhi", etc. may also be created.

S108 creates a "player" relationship between the entity "Gao forest" and the entity "Hengda", inherited from the "player" relationship between the concept "player" and the concept "team".

Similarly, corresponding relationships are created for the remaining player entities.

S109 adds one of the attributes "weight" of the entity "Gao forest". The attributes of an entity inherit from the attributes of a concept.

It is noted that the attributes that an entity can add are limited to the attributes of the concept layer that it inherits. For example, in fig. 1, the attributes of "player" in the concept layer are "native", "weight", "height", "number", "position", and the entities "zhuyijing", "Gao lin", "zheng zhi", "cautiodon jie" corresponding to "player" in the entity layer can only add these attributes. In fig. 1, there is no attribute in "team" in the concept layer, and there is no attribute in "port of entry", "national security", "constant size" of the entity inherited from "team" in the entity layer. In the current knowledge graph construction scheme, for any attribute, the attribute value can be null, for example, all attributes of the entity "high mingen" are null; however, if a new attribute is added to an entity, for example, the attribute "age" is added to "yexi" in the concept layer, the attribute corresponding to "player" in the concept layer must be changed first, and the attribute "age" is added to the attribute corresponding to "player" in the concept layer, otherwise, the attribute "age" cannot be added to "yexi".

The process is as follows: at present, when a knowledge graph is constructed, a constructor is required to have strong abstract capability and know the whole knowledge in the service field, and only then, a large amount of knowledge can be extracted, upper-layer concepts and attributes and the relation between the concepts are created to form templates, and then specific entities are created according to the templates, and corresponding attribute contents are filled, so that the knowledge requirement of the constructor of the knowledge graph is high. In addition, because the construction of the entity layer must be inherited to the template of the concept layer, it is often difficult for constructors to consider the comprehensiveness at the initial stage of construction of the knowledge graph, which results in that when the attributes of a certain entity or the relationship between entities are changed at the later stage, the concept layer must be modified first, which is very troublesome and complicated. For example, a plurality of entities exist in a system, inherit the same concept, and now to add a certain attribute to one or a part of the entities, the attribute must be added to the inherited concept in the concept layer, which is a tedious process.

Disclosure of Invention

The embodiment of the disclosure provides a method, a device, equipment and a storage medium for knowledge graph construction and intelligent response, and mainly solves the technical problems that: at present, when a knowledge graph is constructed, a constructor is required to have strong abstract capability, and the integral knowledge in the service field is required to be known, so that the requirement on the constructor is high; in addition, the construction of the entity layer must be inherited to the template of the concept layer, which is not beneficial to the construction and adjustment of the knowledge map.

In order to solve the above technical problem, an embodiment of the present disclosure provides a method for constructing a knowledge graph, including:

creating entities and adding the relationship among the entities; the relationship among the entities comprises a direct relationship and an indirect relationship; the direct relation is a direct relation between two entities, and the indirect relation is a relation between the two entities which realizes the relation through other entities;

and adding corresponding attribute information for each entity according to the requirement of each entity.

In addition, the embodiment of the present disclosure further provides an intelligent response method adapted to the constructed knowledge graph, including:

acquiring input question information;

processing the problem information to obtain an entity and attribute information corresponding to the problem information;

inquiring in a corresponding knowledge graph according to the entity and the attribute information;

and outputting the inquired result.

The embodiment of the present disclosure further provides a knowledge graph constructing apparatus, including:

an entity creating module for creating an entity;

the relationship adding module is used for adding the relationship among the entities; the relationship among the entities comprises a direct relationship and an indirect relationship; the direct relation is a direct relation between two entities, and the indirect relation is a relation between the two entities which realizes the relation through other entities;

and the attribute adding module is used for adding corresponding attribute information to each entity according to the requirement of each entity.

The embodiment of the present disclosure further provides an intelligent response device, including:

the input module is used for acquiring input question information;

the processing module is used for processing the problem information to obtain an entity and attribute information corresponding to the problem information;

the query module is used for querying in a corresponding knowledge graph according to the entity and the attribute information;

and the output module is used for outputting the inquired result.

The embodiment of the present disclosure further provides a knowledge graph constructing apparatus, including: a first processor, a first memory, and a first communication bus;

the first communication bus is used for realizing connection communication between the first processor and the first memory;

the first processor is configured to execute one or more first programs stored in the first memory to implement the steps of the above-described knowledge graph construction method.

The embodiment of the present disclosure further provides an intelligent answering device, including: a second processor, a second memory, and a second communication bus;

the second communication bus is used for realizing connection communication between the second processor and the second memory;

the second processor is configured to execute one or more second programs stored in the second memory to implement the steps of the intelligent response method described above.

The embodiment of the present disclosure also provides a storage medium, in which computer-executable instructions are stored, and the computer-executable instructions are used for executing the aforementioned knowledge graph construction method or intelligent response method.

The beneficial effects of this disclosure are:

according to the knowledge graph construction and intelligent response method, device, equipment and storage medium provided by the embodiment of the disclosure, the entity is created, the relationship among the entities is added, and the corresponding attribute information is added to each entity according to the requirement of each entity. That is, in the embodiment of the present disclosure, the entity can be directly created, and the relationship and attribute information between the entities is directly added on the basis of the entity, so that the whole creation process is not limited by the concept layer, the degree of freedom is high, and the construction and adjustment of the whole knowledge graph are very convenient. For example, when new attribute information is added to a created entity, the new attribute information can be directly added without being limited by a concept layer. In addition, in the embodiment of the disclosure, the entity is directly created, so that there is no requirement for refining the upper-layer concepts and attributes, which reduces the knowledge requirement for the knowledge graph builder, and even if the builder does not have strong abstraction ability, a better knowledge graph can be constructed, which has better universality.

In addition, the intelligent response method provided in the embodiment of the present disclosure obtains the input question information, processes the question information to obtain the entity and attribute information corresponding to the question information, further queries in the corresponding knowledge graph according to the entity and attribute information, and finally outputs the queried result. Therefore, an intelligent response process adaptive to the constructed knowledge graph is provided, and the effective application of the constructed knowledge graph in the intelligent question and answer field is ensured.

Drawings

FIG. 1 is a prior art knowledge graph constructed in the present manner according to the present disclosure;

FIG. 2 is a schematic basic flowchart of a knowledge graph construction method according to an embodiment of the present disclosure;

fig. 3 is a diagram illustrating the effect of a knowledge map in the field of soccer provided in an embodiment of the present disclosure;

FIG. 4 is a diagram illustrating an effect of a knowledge-map in another football field according to an embodiment of the present disclosure;

fig. 5 is a schematic diagram of a dynamic adjustment process according to an embodiment of the disclosure;

fig. 6 is a schematic basic flow chart of an intelligent response method according to an embodiment of the present disclosure;

fig. 7 is a schematic view of a knowledge graph construction process in the football field according to a second embodiment of the present disclosure;

FIG. 8 is a schematic diagram illustrating a construction process of a knowledge graph in the financial field according to a second embodiment of the disclosure;

FIG. 9 is a diagram of a knowledge-graph effect in the financial field according to a second embodiment of the disclosure;

fig. 10 is a schematic diagram of a process for constructing a knowledge graph in the insurance field according to a second embodiment of the disclosure;

FIG. 11 is a diagram of a knowledge graph effect in the insurance field according to a second embodiment of the present disclosure;

fig. 12 is a schematic structural diagram of an intelligent question answering system provided in the second embodiment of the present disclosure;

fig. 13 is a schematic structural diagram of a knowledge graph constructing apparatus according to a third embodiment of the present disclosure;

fig. 14 is a schematic structural diagram of an intelligent response device provided in a third embodiment of the present disclosure;

FIG. 15 is a schematic structural diagram of a knowledge graph constructing apparatus according to a fourth embodiment of the present disclosure;

fig. 16 is a schematic structural diagram of an intelligent answering device according to a fourth embodiment of the present disclosure.

Detailed Description

Various embodiments of the disclosed concept will now be described in more detail with reference to the accompanying drawings. The disclosed concept may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosed concept to those skilled in the art. Throughout the above description and drawings, the same reference numbers and designations represent the same or similar elements.

It will be understood that, although the terms first, second, etc. may be used herein to describe various elements or operations, these elements or operations should not be limited by these terms. These terms are only used to distinguish one element or operation from another. For example, a first entity may be referred to as a second entity, and similarly, a second entity may be referred to as a first entity without departing from the teachings of the present disclosure.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosed concept. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, regions, portions, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, regions, portions, steps, operations, elements, components, and/or groups thereof.

Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and/or the present disclosure, and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.

The embodiments of the present disclosure will be described in further detail below with reference to the accompanying drawings.

The first embodiment is as follows:

referring to fig. 2, fig. 2 is a basic flow diagram of a knowledge graph constructing method according to an embodiment of the present disclosure, including:

s201: creating an entity;

it should be noted that the entity created in this embodiment may be something that is distinguishable and independent, such as a person, an object, etc., as described in the background. However, for different entities, because their relationships are different, it is possible that the relationships that the different entities have are dependencies. For example, in fig. 1, the entity "Gao forest" and the entity "constant" is essentially an affiliation, except that the affiliation can be more specifically defined as a "player" relationship (essentially, an affiliation between a team and a player). Accordingly, since the relationship between different entities may be a dependency relationship, the entity that is the "master" may also be a refined concept. For example, for "communication equipment company" and "zhongxing communication", wherein the entity "zhongxing communication" is a specific company, is an entity defined in the background art, and is subordinate to "communication equipment company", and for "communication equipment company", it is a collection of companies, corresponding to the concept in the background art. However, in this embodiment, it may be created as an entity, and an inheritance relationship between the concept and the entity in the background art is created as an affiliation relationship between the entities. That is, in the present embodiment, the entity includes not only something that is distinguishable and exists independently, such as a person, an article, and the like, but also a concept derived from the generalization of the entity. Whether a concept needs to be created as an entity should be determined according to the needs of the actual service.

In this embodiment, generally, the created entity should be greater than or equal to two.

S202: adding the relationship among the entities;

in the present embodiment, the relationship between the entities includes a direct relationship and an indirect relationship. Wherein, the direct relation means: direct relationships between two entities, such as "Gao forest" and "constant big" in FIG. 3 are direct relationships; the indirect relationship means: the relationship between two entities is realized by other entities, for example, "Gao forest" and "zheng zhi" in fig. 3 are connected by "constant large", and "Gao forest" and "zheng zhi" are indirect relationships.

In this embodiment, only dependencies between entities may be established, for example, both "Gao forest" and "zhengzhi" in fig. 3 only add dependencies "players" with "constant", but not equal relations "teammates" between "Gao forest" and "zhengzhi". The teammate relationship between "Gao forest" and "zheng zhi" is now determined indirectly by the "player" relationship between both and the entity "constant". In the method, the adding quantity of the relations can be reduced, the complexity of the established knowledge graph is simplified to a certain extent, and meanwhile, the storage space is saved.

However, in this embodiment, various relationships between the entities may be added according to the requirement, for example, referring to the relationship between the entities "national security", "hong kong", "Hengda" shown in fig. 4, i.e. adding "competition" relationship, and the relationship between "Gao forest" and "Zhengzhi" relationship, i.e. adding "teammate" relationship.

S203: and adding corresponding attribute information for each entity according to the requirement of each entity.

In this embodiment, the attribute information that each entity needs to add may be different, which may be controlled by the actual needs of each entity in the service. For example, for an entity "meixi", it may be desirable to add not only basic information such as his height and weight, but also attribute information such as his event, goal number, etc. for a business, and for an entity such as "high mingje", it may be desirable to add only basic information such as his height and weight.

In this embodiment, the relationship between the entities and the attribute information corresponding to the entities may be added after the complete entity is created. However, the relationship between the entities and/or the attribute information of the entities may be correspondingly added in the process of creating the entities. For example, in the knowledge graph shown in fig. 3, after the entity "national security" and the entity "philosophy" are created, the relationship "player" between "national security" and "philosophy" may be added, and the attribute information of "philosophy" may be added. Then, entities such as 'Hengda', 'Zhengzhi' and the like are created.

It should be understood that, for the apparatus for constructing the knowledge graph, certain steps (such as creating entities, adding relationships, adding attribute information, etc.) performed by the apparatus are implemented by corresponding instructions inside the apparatus. Therefore, in this embodiment, the device necessarily generates a corresponding entity creation instruction, a relationship addition instruction, an attribute addition instruction, and the like, and the device implements the corresponding entity creation and the addition of the relationship and the attribute according to these instructions.

Generally, the instruction may be generated by the device according to information crawled by the device, but the instruction may also be generated according to human-computer interaction operation received by the device. That is, in the present embodiment, the apparatus may automatically construct the knowledge-graph by crawling target knowledge for constructing the knowledge-graph, but may also manually construct the knowledge-graph by a person.

When the equipment automatically constructs the knowledge graph by crawling the target knowledge for constructing the knowledge graph, an engineer can set corresponding requirements in the equipment, for example, the requirements of 'teams and players in the football field' are set, and the equipment can automatically crawl the target knowledge meeting the requirements from the Internet through an automatic tool.

It should also be understood that when crawling to the target knowledge, the device should process the target knowledge to obtain the entities in the target knowledge, the relationships among the entities, and the attribute information of the entities, so as to be used for constructing the knowledge graph.

It should be noted that, in this embodiment, after the target knowledge is processed to obtain all entities in the target knowledge, relationships among the entities, and attribute information of the entities, a knowledge graph may be constructed through the process shown in fig. 2.

In this embodiment, in the process of processing the target knowledge to obtain the entities, the relationships between the entities, and the attribute information of the entities in the target knowledge, the knowledge graph may be constructed according to the obtained entities, the relationships between the entities, and the attribute information of the entities through the process shown in fig. 2, so as to speed up the completion speed of construction of the knowledge graph.

In actual business, the attribute information needed is in most cases the attribute information of the entity at the end of the dependency relationship. It should be noted that the entities at the end of the dependency relationship refer to the entities at the lowest level of the whole relationship network in the whole dependency relationship, such as the entities "open philosophy", "Gao forest", "zheng zhi", "caucasian" in fig. 3. Therefore, in a specific implementation manner of this embodiment, corresponding attribute information may be added only to each entity at the end of the relationship, thereby reducing the complexity of the constructed knowledge graph and saving the storage space. It should be noted that, during the addition, the addition needs to be performed according to the needs of each entity, for example, as shown in fig. 3, attributes are added for "zhangzheng", "Gao lin" and "zheng zhi", and no attribute is added for "shangxijg".

It should be understood that, since knowledge is continuously updated, as target knowledge used for constructing the knowledge graph changes, entities that have already been created, relationships between entities, and attributes of entities may change, and in this case, adjustment needs to be performed dynamically, which may be specifically shown in fig. 5, including:

s501: after the entities are created and the relationships among the entities are added, if a new entity which is not created exists, determining the entity directly related to the new entity;

s502: when the entity directly associated with the new entity is created, the relationship between the new entity and the entity directly associated with the new entity is added, and corresponding attribute information is added to the new entity according to the requirement of the new entity.

It should be understood that the relationships in the knowledge graph may be directly altered as the relationships between the entities change. For example, taking fig. 3 as an example, assuming that "zhanghui" leaves "national ampere" to "heng", in fig. 3, it is necessary to disconnect "zhanghui" and "national ampere", establish the connection between "zhanghui" and "heng", and add the connection relationship as "player".

It should be understood that when the attribute information of an entity changes, the corresponding attribute information may be directly changed. For example, in fig. 3, assuming that the weight of the "thin philosophy" is increased from 69 to 80, the attribute value "69" in the weight attribute needs to be updated to "80" in fig. 3.

It should be understood that the addition of attribute information to an entity may be straightforward. For example, in fig. 3, if it is assumed that the attribute "age" needs to be added to "thin philosophy", in fig. 3, the attribute "age" may be directly created and an age value may be added.

It should also be understood that when some entity or attribute information needs to be removed, it can be deleted directly in the established knowledge graph.

It should be noted that, when the created entities, the relationships between the entities, and the attributes of the entities change, the process of dynamically adjusting may be automatically performed by the device, but may also be manually updated.

In this embodiment, at least two specific construction modes exist when the knowledge graph is constructed:

one is as follows: a main entity can be created first, then a first sub-entity is created under the main entity, and the relationship between the main entity and the first sub-entity is added; continuing to create second secondary entities under each first secondary entity, and adding the relationship between the first secondary entities and the second secondary entities; …, creating the nth order entity under each nth-1 order entity, and adding the relationship between the nth-1 order entity and the nth order entity. In this embodiment, n should be 2 or more.

I.e. entities are created in order from top to bottom in dependency relationship. For example, the knowledge graph shown in fig. 3 is created by first creating a main entity "team", then creating first entities "port", "national security", and "constant", and then creating second entities "zhuangzhi", "Gao forest", "zheng zhi", and "cautious jid" under each first entity.

It should be understood that the nth order entity may not be present in this embodiment. I.e. the knowledge-graph comprises only the primary entity and the first secondary entity. For example, as shown in FIG. 4, "Zhang Zheng", "Gao Ling", "Zheng Zhi", "Gaogong Jie" is the first entity, and "Shang gang", "Guan" and "Heng Da" are the main entities.

It should be noted that, in this embodiment, the primary entity refers to an entity located at the uppermost layer in the dependency relationship, and the primary entity is an entity directly subordinate to the primary entity. Correspondingly, the second sub-entity is an entity directly subordinate to the first sub-entity, and the nth sub-entity is an entity directly subordinate to the (n-1) th sub-entity.

The second step is as follows: a first entity can be created, a second entity directly related to the first entity can be created, and the relationship between the first entity and the second entity can be added; when the m-th entity has directly associated entities besides the m-1 entity, creating an m + 1-th entity according to the m-th entity which has directly associated entities besides the m-1 entity, and adding the relationship between the m-th entity and the m + 1-th entity. In this embodiment, m should be 2 or more.

That is, one entity can be created arbitrarily, and epidemic creation is performed by using the association relationship of the entity as a link based on the created entity.

For example, taking the knowledge graph shown in fig. 4 as an example, an entity may be created at will, and if the created entity is "national security", the entity directly associated with "national security" is found to have a player "zhangzheng", and two competitors "shang gang" and "heng", and then entities "zhangzheng zheng", "shang gang" and "heng gang" connected to "national security" are created. Further, detecting the Zhang Zheng discovers that no entity directly related to the Zhang Zheng except for the national security exists, ends the establishment line of the entity Zhang Zheng, detects the Shanggang discovers that the entity directly related to the Zhang Zheng except for the national security exists in GaoShi, and establishes the entity GaoShi connected with the Shanggang; detecting "Hengda" finds out that the entities directly related to "Gao forest" and "Zhengzhi" besides "Guoan", and creates entities "Gao forest" and "Zhengzhi" connected with "Hengda"; detecting entities 'GaoShijie', 'Gao Lin' and 'Zhengzhi', finding that other directly associated entities do not exist, and finishing the entity creation.

In the embodiment, the two entity creating processes can well create the entities required by the knowledge graph, and the method has good universality. However, this does not represent that an entity can be created in only the above two ways in this embodiment. In fact, any manner in which the required entities of a knowledge graph can be created is suitable for the present implementation and is intended to be within the scope of the embodiments of the present disclosure.

It should be understood that, in the embodiment, the relevant data of the knowledge-graph should be stored during the process of constructing the knowledge-graph or after the knowledge-graph is constructed.

In addition, the embodiment also provides an intelligent response method which can be matched with the created knowledge graph and can realize intelligent response to the question. As shown in fig. 6, the method comprises the following steps:

s601: acquiring input question information;

in this embodiment, the input mode of the question information includes, but is not limited to, wechat access, ASR (automatic speech Recognition) speech input, text input by an application or a web interface, and the like.

S602: processing the problem information to obtain an entity and attribute information corresponding to the problem information;

in this embodiment, processing the question information includes converting the question information into text information (text input for the matter, i.e., conversion is not required). And preprocessing the problem text, wherein the preprocessing comprises word segmentation, error correction, part of speech tagging, normalization processing and the like. It should be understood that the foregoing part-of-speech tagging refers to the property of tagging the subject, predicate, object, state, etc. of a word in a sentence. The normalization process is to unify the words. For example, if the user inputs "Zhongxing" and the knowledge graph records "Zhongxing communication", the normalization will process "Zhongxing" as "Zhongxing communication".

And then, preprocessing the problem text, performing semantic analysis on the preprocessed content, and extracting entity and attribute information contained in the preprocessed content by combining a knowledge graph.

S603: inquiring in a corresponding knowledge graph according to the entity and attribute information corresponding to the problem information;

for example, for the knowledge-graph shown in fig. 3, let the user input be "how much zheng is there? "yes", the extracted entity is "zheng zhi", attribute is "weight", and the result "75" can be found out in fig. 3.

It should be noted that, if the problem information is processed, the entity and the attribute information in the problem information can be obtained, that is, the entity and the attribute information in the problem information are directly used to perform the query in the corresponding knowledge graph.

If the problem information is processed to find that only the entity or attribute information exists in the problem information, the completion can be performed by acquiring the entity or attribute adopted before. Specifically, the method comprises the following steps:

if the problem information is processed to obtain only the entity in the problem information, the attribute information used in the last query is obtained, and the attribute information is used as the attribute information corresponding to the problem information to be queried. For example, for the knowledge-graph shown in fig. 3, let us say that the user has input the question "how much zheng is there? "the entity obtained is" Zhengzhi "and the attribute is" weight "; what the user input at this query is "thin philosophy? "this time, only the entity" Zhang Zun Zhen "can be obtained, and at this time, the attribute" weight "of the last query is taken as the attribute of the current query, so that the result of" 69 "can be obtained.

If the problem information is processed to obtain only the attribute in the problem information, an entity used in the last query is obtained, and the entity is used as the entity corresponding to the problem information to be queried. For example, for the knowledge-graph shown in fig. 3, let us say that the user has input the question "how much zheng is there? "the entity obtained is" Zhengzhi "and the attribute is" weight "; what the user entered at this query was "how much height? "this time, only attribute" height "can be obtained, at this time, entity" zhengzhi "at last time of inquiry is regarded as entity at this time of inquiry, and then result" 180 "can be obtained.

Therefore, context omission and recovery can be carried out on the basis of the knowledge graph, the context omission and recovery is realized according to the characteristic that a subject or intention is usually omitted in language habits, the subject and the intention correspond to the entity and the attribute in the knowledge graph, if one of the subject and the intention is lacked, the corresponding data extracted last time is used for replacing the subject and the intention, the accuracy of intelligent question answering can be effectively improved, and the whole accuracy is limited only by the word segmentation accuracy theoretically.

In this embodiment, the device may generate a corresponding SQL (structured query Language) query statement (or SPARQL (Simple Protocol and RDFQuery Language) query statement) according to the extracted entity, attribute, and the like, and query the map data stored in the data database (or in an RDF (Resource Description Framework) file) to obtain an answer.

It should be particularly noted that the knowledge graph applied in this embodiment may be a knowledge graph constructed by the above-mentioned knowledge graph construction method.

S604: and outputting the inquired result.

In this embodiment, the answer may be output to the user in a manner that the answer is played at the terminal after being pushed to a WeChat or TTS (Track & Trace system, two-way tracing system) text is converted into speech through a WeChat interface, and the text is displayed on an application or web interface.

According to the knowledge graph construction method provided by the embodiment of the disclosure, the entity is created, the relationship among the entities is added, and the corresponding attribute information is added to each entity according to the requirement of each entity. That is, in the embodiment of the present disclosure, the entity can be directly created, and the relationship and attribute information between the entities is directly added on the basis of the entity, so that the whole creation process is not limited by the concept layer, the degree of freedom is high, and the construction and adjustment of the whole knowledge graph are very convenient. For example, when new attribute information is added to a created entity, the new attribute information can be directly added without being limited by a concept layer. In addition, in the embodiment of the disclosure, the entity is directly created, so that there is no requirement for refining the upper-layer concepts and attributes, which reduces the knowledge requirement for the knowledge graph builder, and even if the builder does not have strong abstraction ability, a better knowledge graph can be constructed, which has better universality.

In addition, according to the intelligent response method provided by this embodiment, the entity and attribute information corresponding to the question information are obtained by acquiring the input question information and processing the question information, and then the query is performed in the corresponding knowledge graph according to the entity and attribute information, and finally the queried result is output. Therefore, an intelligent response process adaptive to the constructed knowledge graph is provided, and the effective application of the constructed knowledge graph in the intelligent question and answer field is ensured. In addition, in the embodiment, the context can be omitted and recovered based on the knowledge graph, and the data of the query is supplemented by the corresponding data in the knowledge graph in the last query, so that the method is more consistent with language habits, and the semantic accuracy of recovery is effectively improved.

Example two:

this embodiment further illustrates the scheme of the present disclosure through several more specific knowledge graph construction processes based on the first embodiment.

Referring to fig. 7, fig. 7 is a schematic diagram of a knowledge graph construction process in the football field provided in this embodiment, and includes:

s701: creating an entity "team";

s702: creating entities "hong Kong", "Guoan", "Hengda";

s703: adding the relationship among 'Shanghao', 'Guoan', 'Hengda' and 'team';

in this embodiment, the relationship between "hong kong", "national security", "constant big", and "team" is the relationship of "is-a". It should be noted that in the knowledge representation field, is-a refers to the parent-child inheritance relationship of a class, for example, that class D is a child of another class B (class B is a parent of class D).

S704: entities "Zhang Dichen", "Gao forest", "Zheng Zhi", "GaoZhi Jie" are created.

S705: adding a relation of 'player' between an entity 'Zhanghui' and an entity 'Guoan'; adding the 'constant' relationship between the entity 'Gao forest', 'Zhengzhi' and the entity 'Guoan'; adding a relation of 'player' between the entity 'GaoShigjie' and the entity 'Shanggang';

s706: adding the attributes of the entities 'Zhangling philosophy', 'Gao forest' and 'Zhengzhi' and filling in the attribute values.

In this embodiment, the attributes of the entities "Zhang Diji", "Gao forest" and "Zheng Zhi" are "native", "weight", "height", "number" and "position". Wherein, the attribute values of the entity Zhang Ding Zhe are as follows in sequence: "Hubei", "69", "180", "10", "foreward"; the attribute values of the entity "Gao forest" are, in order: "Shenyang", "75", "180", "10", "defensive", "Shenyang", "75", "180", "10", "Shenyang", "Shiyang",; the attribute values of entity 'zheng zhi' are as follows: "Zhengzhou", "75", "180", "29", "front".

The resulting map effect graph can be seen in fig. 3.

Referring to fig. 8, fig. 8 is a schematic diagram of a knowledge graph construction process in the financial field provided in this embodiment, and includes:

s801: creating an entity "financing product";

s802: creating entities "shared", "private", "honorable";

s803: adding the relationship among sharing type, private sharing type, honorable sharing type and financing product;

in the present embodiment, the relationship between the "shared type", "private type", "honorable type", and "financing product" is the relationship of "is-a".

S804: attributes of entities "shared", "private", "honorable" are added and attribute values are filled in.

In this embodiment, the attributes of "shared", "private", and "honorable" are "risk", "purchase amount", and "profitability". Wherein, the attribute values of the entity 'sharing type' are as follows in sequence: "Low", "5 ten thousand", "4.65"; the attribute values of the entity 'private type' are as follows in sequence: "middle", "20 ten thousand", "5.83"; the attribute values of the entity "respect type" are as follows in sequence: "high", "100 ten thousand", "9.08".

The resulting map effect graph can be seen in fig. 9.

Referring to fig. 10, fig. 10 is a schematic diagram of a process for constructing a knowledge graph in the insurance field provided in this embodiment, and includes:

s1001: creating an entity "insurance category";

s1002: creating entity "A Risk", "B Risk";

s1003: adding the relationship among the risk A, the risk B and the insurance category;

in the present embodiment, the relationship between "risk a", "risk B", and "insurance type" is the relationship of "is-a".

S1004: adding the attributes of the entities 'A risk' and 'B risk' and filling in the attribute values.

In this embodiment, the attributes of the "risk a" and the "risk B" are "type", "years of payment", "insured person" and "annual premium". Wherein, the attribute values of the entity 'A risk' are as follows in sequence: "red type", "10 years", "infant" and "5000"; the attribute values of the entity 'B risk' are as follows in sequence: "basic type", "1 year", "adult", "200".

The resulting map effect graph can be seen in fig. 11.

On the basis of the first embodiment, the present embodiment further illustrates the scheme of the present disclosure by a more specific intelligent question-answering system.

Referring to fig. 12, the intelligent question answering system structure includes: the system comprises a map storage module 1201, a map management module 1202, an input and output module 1203, a preprocessing module 1204, an intention identification module 1205 and a map query module 1206. Wherein:

the map storage module 1201 realizes storage of map data, and the storage mode supports a database or an RDF format file.

The map management module 1202 implements management and maintenance functions of map data, and may use an automated tool to crawl a structured knowledge form from the internet, or provide a management and maintenance interface to perform manual management, and store data in the map storage module 1201.

The input/output module 1203 receives the questions input by the user, and outputs the answers of the system to the user. Supports Wechat access, voice access (realized by ASR voice to text and TTS text to voice) and user text interaction interface.

The preprocessing module 1204 receives the user question transmitted from the input/output 1203 module, performs word segmentation, error correction, part of speech tagging, normalization, and the like on the user question, and transmits the processing result to the intention recognition module 1205 for further processing.

The intention recognition module 1205 recognizes the intention of the preprocessed content in combination with the knowledge graph, and extracts elements including entities, attributes and parameters contained therein. If the attributes are extracted at this time but the entities are not extracted, the attributes are replaced by the entities extracted at the last time, and if the entities are extracted at this time but the attributes are not extracted, the attributes extracted at the last time are replaced by the attributes extracted at the last time, so that the context omitting and recovering function based on the knowledge graph is realized.

The map query module 1206 generates a corresponding SQL query statement (or SPARQL query statement) according to the extracted entities, attributes and parameters, and queries the map storage module 1201 to obtain an answer.

According to the process and the system provided by the embodiment, it can be seen that the knowledge graph construction method and the intelligent response method provided by the embodiment of the disclosure have at least the following advantages:

1. the knowledge graph construction method provided by the embodiment of the disclosure removes the concept layer of the traditional knowledge graph, reduces the skill requirement on personnel, and improves the construction and maintenance efficiency of the knowledge graph;

2. the intelligent answering method provided by the embodiment of the disclosure can effectively combine the knowledge graph to realize intelligent question answering and improve the intelligent degree.

3. The intelligent response method provided by the embodiment of the disclosure can perform context omission recovery based on the knowledge graph, and can effectively improve the semantic accuracy of recovery.

Example three:

referring to fig. 13, fig. 13 is a knowledge graph constructing apparatus 13 according to a third embodiment of the present disclosure, including: an entity creation module 131, a relationship addition module 132, and an attribute addition module 133. Wherein:

an entity creation module 131 for creating an entity;

a relationship adding module 132, configured to add relationships between entities;

the attribute adding module 133 is configured to add corresponding attribute information to each entity according to the needs of each entity.

In this embodiment, the concept can be created as an entity, and the inheritance relationship between the concept and the entity in the background art can be created as an affiliation relationship between the entities. Meanwhile, the entity created in this embodiment may be something that is distinguishable and exists independently, such as a person, an article, and the like, as described in the background.

In this embodiment, whether a concept needs to be created as an entity should be determined according to the needs of the actual service.

In the present embodiment, the relationship between the entities includes a direct relationship and an indirect relationship. Wherein, the direct relation means: a direct association relationship between two entities; the indirect relationship means: and the two entities realize the associated relation through other entities.

In this embodiment, the attribute information that each entity needs to add may be different, which may be controlled by the actual needs of each entity in the service.

In this embodiment, the relationship between the entities and the attribute information corresponding to the entities may be added after the complete entity is created. However, the relationship between the entities and/or the attribute information of the entities may be correspondingly added in the process of creating the entities.

It should be understood that for the knowledge-graph constructing apparatus for constructing the knowledge-graph, the creation of entities, the addition of relationships, the addition of attribute information, and the like are all realized by corresponding instructions therein. Therefore, in this embodiment, the knowledge-graph constructing apparatus inevitably generates a corresponding entity creating instruction, a relationship adding instruction, an attribute adding instruction, and the like, and the knowledge-graph constructing apparatus implements the corresponding entity creating and the adding of the relationship and the attribute according to these instructions.

Generally, the instruction may be generated by the knowledge graph building apparatus according to information crawled by the knowledge graph building apparatus, but the instruction may also be generated according to human-computer interaction operation received by the device. That is, in the present embodiment, the knowledge-graph constructing means may automatically construct the knowledge graph by crawling the target knowledge for constructing the knowledge graph, but may manually construct the knowledge graph by a human.

When the knowledge graph constructing device automatically constructs the knowledge graph by crawling the target knowledge for constructing the knowledge graph, an engineer can set corresponding requirements in the knowledge graph constructing device, for example, the requirements of teams and players in the football field are set, and the knowledge graph constructing device can automatically crawl the target knowledge meeting the requirements from the Internet through an automatic tool.

It should also be understood that, when crawling to the target knowledge, the knowledge graph constructing apparatus should process the target knowledge to obtain the entities in the target knowledge, the relationships among the entities, and the attribute information of the entities, so as to be used for constructing the knowledge graph.

It should be noted that, in this embodiment, after the target knowledge is processed, all entities in the target knowledge, relationships between the entities, and attribute information of the entities are obtained, and then a knowledge graph may be constructed.

In this embodiment, in the process of processing the target knowledge to obtain the entities, the relationships among the entities, and the attribute information of the entities in the target knowledge, the knowledge graph may be constructed according to the obtained entities, the relationships among the entities, and the attribute information of the entities, so as to accelerate the speed of constructing the knowledge graph.

In actual business, the attribute information needed is in most cases the attribute information of the entity at the end of the dependency relationship. It should be noted that the entity at the end of the dependency relationship refers to each entity at the lowest level of the entire relationship network in the entire dependency relationship, and therefore, in a specific implementation manner of this embodiment, corresponding attribute information may be added only to each entity at the end of the relationship, so that complexity of the constructed knowledge graph is reduced, and a storage space is saved. It should be noted that, at the time of addition, addition is still required according to the needs of each entity.

It should be understood that, since knowledge is continuously updated, as target knowledge used for constructing the knowledge graph changes, entities that have already been created, relationships between entities, and attributes of entities may change, and in this case, dynamic adjustment is required, specifically:

after the entities are created and the relationships among the entities are added, if a new entity which is not created exists, the entity directly associated with the new entity is determined, when the entity directly associated with the new entity is created, the new entity is created and the relationship between the new entity and the entity directly associated with the new entity is added, and corresponding attribute information is added to the new entity according to the requirements of the new entity.

It should be understood that the relationships in the knowledge graph may be directly altered as the relationships between the entities change.

It should be understood that when the attribute information of an entity changes, the corresponding attribute information may be directly changed.

It should be understood that the addition of attribute information to an entity may be straightforward.

It should be noted that, when the created entities, the relationships between the entities, and the attributes of the entities change, the process of dynamically adjusting may be automatically performed by the knowledge graph building apparatus, but may also be manually updated.

In this embodiment, when constructing the knowledge graph, the entity creating module and the relationship adding module may construct the entity and add the relationship at least in the following two specific construction manners:

one is as follows: the entity creation module may create a primary entity first and a first secondary entity under the primary entity, and create an nth secondary entity under the (n-1) th secondary entity when the nth secondary entity exists. n is 2 or more.

The relationship adding module may add a relationship between the primary entity and the first secondary entity, and after creating the nth secondary entity, add a relationship between the n-1 th secondary entity and the nth secondary entity.

One is as follows: the entity creating module may first create a first entity, create a second entity directly associated with the first entity, and create an m +1 th entity according to the fact that the m-th entity has directly associated entities in addition to the m-1 th entity when the m-th entity has directly associated entities in addition to the m-1 th entity. m is greater than or equal to 2.

The relationship adding module may add a relationship between the first entity and the second entity, and add a relationship between the m-th entity and the m + 1-th entity after the m + 1-th entity is created.

In addition, the embodiment also provides an intelligent answering device which can be matched with the created knowledge graph and can realize intelligent answering of questions. Referring to fig. 14, smart responder 14 includes: an input module 141, a processing module 142, a query module 143, and an output module 144. Wherein,

an input module 141 for acquiring input question information;

the processing module 142 is configured to process the question information to obtain an entity and attribute information corresponding to the question information;

the query module 143 is configured to query in the corresponding knowledge graph according to the entity and the attribute information;

and the output module 144 is used for outputting the queried result.

Wherein, the processing module is specifically configured to: processing the problem information to obtain entity and attribute information in the problem information; if the problem information is processed to obtain only the entity in the problem information, acquiring the attribute information used in the last query, and taking the attribute information as the attribute information corresponding to the problem information; and if the problem information is processed to obtain only the attribute information in the problem information, acquiring an entity used in the last query, and taking the entity as the entity corresponding to the problem information.

According to the knowledge graph constructing device provided by the embodiment of the disclosure, the relationship among the entities is added by creating the entities, and the corresponding attribute information is added for each entity according to the requirement of each entity. That is, in the embodiment of the present disclosure, the entity can be directly created, and the relationship and attribute information between the entities is directly added on the basis of the entity, so that the whole creation process is not limited by the concept layer, the degree of freedom is high, and the construction and adjustment of the whole knowledge graph are very convenient. For example, when new attribute information is added to a created entity, the new attribute information can be directly added without being limited by a concept layer. In addition, in the embodiment of the disclosure, the entity is directly created, so that there is no requirement for refining the upper-layer concepts and attributes, which reduces the knowledge requirement for the knowledge graph builder, and even if the builder does not have strong abstraction ability, a better knowledge graph can be constructed, which has better universality.

In addition, according to the intelligent response device provided by this embodiment, the entity and attribute information corresponding to the question information are obtained by acquiring the input question information and processing the question information, and then the query is performed in the corresponding knowledge graph according to the entity and attribute information, and finally the queried result is output. Therefore, an intelligent response process adaptive to the constructed knowledge graph is provided, and the effective application of the constructed knowledge graph in the intelligent question and answer field is ensured. In addition, in the embodiment, the context can be omitted and recovered based on the knowledge graph, and the data of the query is supplemented by the corresponding data in the knowledge graph in the last query, so that the method is more consistent with language habits, and the semantic accuracy of recovery is effectively improved.

Example four:

the present embodiment provides a knowledge-graph constructing apparatus, as shown in fig. 15, which includes a first processor 151, a first memory 152, and a first communication bus 153. Wherein:

the first communication bus 153 is used for realizing connection communication between the first processor 151 and the first memory 152;

the first processor 151 is configured to execute one or more first programs stored in the first memory 152 to implement the steps of the knowledge graph constructing method according to the first embodiment and/or the second embodiment.

The present embodiment also provides an intelligent answering device, which is shown in fig. 16 and comprises a second processor 161, a second memory 162 and a second communication bus 163. Wherein:

the second communication bus 163 is used for realizing connection communication between the second processor 161 and the second memory 162;

the second processor 161 is configured to execute one or more second programs stored in the second memory 162 to implement the steps of the intelligent response method according to the first embodiment and/or the second embodiment.

The present embodiments also provide a storage medium including volatile or nonvolatile, removable or non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, computer program modules or other data. Storage media includes, but is not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically erasable programmable Read-Only Memory), flash Memory or other Memory technology, CD-ROM (Compact disk Read-Only Memory), Digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by a computer.

The storage medium provided in this embodiment stores computer-executable instructions, which can be executed by one or more processors to implement the steps of the knowledge graph constructing method or the intelligent response method described in the first embodiment and/or the second embodiment. And will not be described in detail herein.

It will be apparent to those skilled in the art that all or some of the steps of the methods, systems, functional modules/units in the devices disclosed above may be implemented as software (which may be implemented in computer program code executable by a computing device), firmware, hardware, and suitable combinations thereof. In a hardware implementation, the division between functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be performed by several physical components in cooperation. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit.

In addition, communication media typically embodies computer readable instructions, data structures, computer program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media as known to one of ordinary skill in the art. Thus, the present invention is not limited to any specific combination of hardware and software.

The foregoing is a more detailed description of embodiments of the present invention, and the present invention is not to be considered limited to such descriptions. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.

Claims

1. A knowledge graph construction method comprises the following steps:

2. The method of knowledge-graph construction according to claim 1, prior to the creating entities and adding relationships between entities, further comprising:

crawling target knowledge for constructing a knowledge graph;

and processing the target knowledge to obtain entities, the relationship among the entities and the attribute information of the entities.

3. The method of knowledge-graph construction according to claim 2 wherein creating entities and adding relationships between entities comprises:

and in the process of processing the target knowledge to obtain the entities, the relationships among the entities and the attribute information of the entities, creating the entities according to the processed entities and the relationships among the entities and adding the relationships among the entities.

4. The method of knowledge-graph construction according to claim 1 wherein there are dependencies between entities;

the adding corresponding attribute information to each entity according to the needs of each entity includes:

adding corresponding attribute information for each entity at the relation end according to the requirement of each entity at the relation end; and the entities at the tail end of the relationship are the entities at the lowest level of the whole relationship network in the dependency relationship.

5. The method of knowledge-graph construction according to claim 1, after creating entities and adding relationships between entities, further comprising:

if the new entity which is not created exists, determining the entity directly related to the new entity;

when the entity directly associated with the new entity is created, the new entity is created and the relationship between the new entity and the entity directly associated with the new entity is added.

6. The method of knowledge-graph construction according to any one of claims 1 to 5 wherein creating entities and adding relationships between entities comprises:

creating a main entity, creating a first sub-entity under the main entity, and adding the relationship between the main entity and the first sub-entity; when an nth-time entity exists, creating the nth-time entity under an (n-1) th-time entity, and adding a relation between the (n-1) th-time entity and the nth-time entity; n is greater than or equal to 2;

or,

creating a first entity, creating a second entity directly associated with the first entity, and adding a relationship between the first entity and the second entity; when the m-th entity has a directly associated entity besides the m-1 entity, creating an m + 1-th entity according to the m-th entity which has a directly associated entity besides the m-1 entity, and adding a relationship between the m-th entity and the m + 1-th entity; and m is greater than or equal to 2.

7. An intelligent answering method, comprising:

acquiring input question information;

and outputting the inquired result.

8. The intelligent answering method of claim 7, wherein the processing the question information to obtain the entity and attribute information corresponding to the question information comprises:

processing the problem information to obtain entity and attribute information in the problem information;

if the problem information is processed to obtain only the entity in the problem information, acquiring the attribute information used in the last query, and taking the attribute information as the attribute information corresponding to the problem information;

and if the problem information is processed to obtain only the attribute information in the problem information, acquiring an entity used in the last query, and taking the entity as the entity corresponding to the problem information.

9. The smart response method according to claim 7 or 8, wherein the knowledge-graph is a knowledge-graph constructed by the knowledge-graph construction method according to any one of claims 1 to 6.

10. A knowledge-graph building apparatus comprising:

an entity creating module for creating an entity;

11. The knowledge-graph constructing apparatus of claim 10,

the entity creation module is specifically configured to: creating a main entity, creating a first sub-entity under the main entity, and creating an n-th sub-entity under an n-1 th sub-entity when the n-th sub-entity exists; n is greater than or equal to 2;

the relationship adding module is specifically configured to: adding the relation between the main entity and the first secondary entity, and after creating the nth secondary entity, adding the relation between the n-1 secondary entity and the nth secondary entity;

or,

the entity creation module is specifically configured to: creating a first entity, creating a second entity directly associated with the first entity, and creating an m +1 th entity according to the fact that the m entity also has directly associated entities except the m-1 th entity when the m entity also has directly associated entities except the m-1 th entity, wherein m is more than or equal to 2;

the relationship adding module is specifically configured to: adding the relationship between the first entity and the second entity, and after creating the m +1 th entity, adding the relationship between the m +1 th entity and the m +1 th entity.

12. An intelligent answering device comprising:

the input module is used for acquiring input question information;

and the output module is used for outputting the inquired result.

13. The intelligent answering device of claim 12,

the processing module is specifically configured to: processing the problem information to obtain entity and attribute information in the problem information;

14. The smart responder according to claim 12 or 13, wherein the knowledge-graph is a knowledge-graph constructed by the knowledge-graph construction method according to any one of claims 1-6.

15. A knowledge-graph building apparatus comprising: a first processor, a first memory, and a first communication bus;

the first processor is configured to execute one or more first programs stored in the first memory to implement the steps of the knowledge-graph construction method of any one of claims 1-6.

16. An intelligent answering machine comprising: a second processor, a second memory, and a second communication bus;

the second processor is configured to execute one or more second programs stored in the second memory to implement the steps of the intelligent response method of any of claims 7-9.

17. A storage medium having stored therein computer-executable instructions for performing the method of knowledge-graph construction according to any one of claims 1-6, or for performing the method of intelligent response according to any one of claims 7-9.