CN109726279B

CN109726279B - Data processing method and device

Info

Publication number: CN109726279B
Application number: CN201811648577.8A
Authority: CN
Inventors: 李让; 胡长建
Original assignee: Lenovo Beijing Ltd
Current assignee: Lenovo Beijing Ltd
Priority date: 2018-12-30
Filing date: 2018-12-30
Publication date: 2021-05-18
Anticipated expiration: 2038-12-30
Also published as: CN109726279A

Abstract

The application discloses a data processing method and device, after input information is obtained, first content matched with the input information and second content associated with the first content are obtained, feedback information is obtained based on the first content and the second content associated with the first content, and the feedback information is output, so that feedback information meeting the expectation of a user is output based on the input information. The first content and the second content are both related to input information, and the input information can be a user problem provided by a user for a user requirement, so that the first content and the second content can be matched with the user requirement, that is, the feedback information obtained based on the first content and the second content can effectively solve the user requirement of the input information of the man-machine conversation, further, the user requirement of the man-machine conversation can be effectively solved, and the user experience is improved.

Description

Data processing method and device

Technical Field

The present application belongs to the field of data processing technologies, and in particular, to a data processing method and apparatus.

Background

With the development of information technology, intelligent customer service can provide more and more abundant online business services for users, such as online intelligent question-answering service, so as to automatically answer user questions (namely questions provided by users) through the online intelligent question-answering service, for example, when the online intelligent question-answering service is called, an online business consultation interface is provided for the users, the user questions are obtained through the online business consultation interface, and answers to the user questions are given after the user questions are analyzed.

If the user questions are: and the screen size of the mobile phone product can be found from a product knowledge base through the online intelligent question-answering service, and then the parameters representing the screen size of the mobile phone product are spliced into an answer template to obtain and output the answer of the user question.

However, when a user asks a question, it may not only be the parameters of the product he wants to obtain, for example, for the user question of the screen size of the product of the mobile phone, the user actually intends to know whether the mobile phone is suitable for being held by one hand, and therefore, if only the answer of the screen size is given, the user needs to search for the data by himself to make a judgment.

Disclosure of Invention

In view of this, an object of the present application is to disclose a data processing method and apparatus, which are used for obtaining feedback information meeting the expected expectations of users based on input information, effectively meeting the user requirements of the current human-computer session, and improving the user experience.

The application discloses a data processing method, which is applied to an intelligent session system, wherein the intelligent session system can respond to received input information and provide feedback information, and the method comprises the following steps:

acquiring input information, wherein the input information is information input by a user in the process of performing man-machine conversation through the intelligent conversation system;

obtaining first content matched with the input information and second content associated with the first content;

obtaining feedback information based on the first content and second content associated with the first content;

and outputting the feedback information.

Preferably, one of the first content and the second content is an attribute value of a first attribute of a first object, and the other is an attribute value of a second attribute of the first object, where the first attribute and the second attribute are different;

or

One of the first content and the second content is an attribute value of a first attribute of a first object, and the other is an attribute value of a first attribute of a second object, and the first object and the second object are different.

Preferably, the obtaining of the first content matching the input information and the second content associated with the first content includes:

obtaining first content matched with the input information from a preset knowledge graph;

and obtaining the content having a specific relation with the first content in the preset knowledge graph, wherein the content having a characteristic relation with the first content is the second content associated with the first content.

Preferably, the obtaining of the content having a specific relationship with the first content in the preset knowledge-graph comprises:

acquiring a first node corresponding to the first content from the preset knowledge graph, wherein the preset knowledge graph records an attribute value of at least one attribute of at least one object, the preset knowledge graph takes object identification information and the attribute value of each attribute of the object as nodes, and the attributes of the object as edge connecting nodes;

acquiring a second node of the first node corresponding to the first content in the preset knowledge graph, wherein the direct connection indicates that the first node and the second node are connected through an edge;

and acquiring a third node which is directly connected with the second node in the preset knowledge graph, and taking the content corresponding to the third node as the content having a specific relation with the first content.

obtaining a first node corresponding to the first content in one of a first knowledge graph and a second knowledge graph in the preset knowledge graph, wherein one of the first knowledge graph and the second knowledge graph records an attribute value of at least one attribute of a first object, the other one records an attribute value of at least one attribute of a second object, each knowledge graph takes object identification information of the respective recorded object and the attribute value of each attribute of the object as nodes, and the attributes of the object as edges to connect the nodes;

obtaining label information of the first node, wherein the label information of the first node is used for indicating the object requirement of first content adaptation;

and obtaining a second node in the other of the first knowledge-graph and the second knowledge-graph based on the label information of the first node, and taking the content corresponding to the second node as the content having a specific relationship with the first content, wherein the label information of the second node and the label information of the first node indicate the same meaning.

Preferably, the method further comprises: connecting the first node and the second node by an edge to construct the first knowledge-graph and the second knowledge-graph into a knowledge-graph.

Preferably, the obtaining feedback information based on the first content and the second content associated with the first content includes:

composing the feedback information from the first content and the second content;

or

And modifying the first content and the second content to obtain the feedback information.

The application also discloses a data processing device, is applied to intelligent conversational system, intelligent conversational system can respond to the input information that receives and provide feedback information, the device includes:

the intelligent conversation system comprises a first obtaining unit, a second obtaining unit and a third obtaining unit, wherein the first obtaining unit is used for obtaining input information, and the input information is information input by a user in a man-machine conversation process through the intelligent conversation system;

a second obtaining unit configured to obtain a first content matching the input information and a second content associated with the first content;

a third obtaining unit configured to obtain feedback information based on the first content and a second content associated with the first content;

and the output unit is used for outputting the feedback information.

or

Preferably, the second obtaining unit includes:

the first acquisition module is used for acquiring first content matched with the input information from a preset knowledge graph;

a second obtaining module, configured to obtain, in the preset knowledge graph, content having a specific relationship with the first content, where the content having a characteristic relationship with the first content is a second content associated with the first content.

Preferably, the second obtaining module includes:

a first obtaining sub-module, configured to obtain a first node corresponding to the first content in the preset knowledge graph, where the preset knowledge graph records an attribute value of at least one attribute of at least one object, and the preset knowledge graph takes object identification information and the attribute value of each attribute of the object as nodes and takes the attribute of the object as an edge connection node;

the second obtaining submodule is used for obtaining a second node which is in direct connection with the first node corresponding to the first content in the preset knowledge graph, and the direct connection indicates that the first node and the second node are connected through an edge;

and the third obtaining submodule is used for obtaining a third node which is directly connected with the second node in the preset knowledge graph, and taking the content corresponding to the third node as the content which has a specific relation with the first content.

Preferably, the second obtaining module includes:

a first obtaining sub-module, configured to obtain a first node corresponding to the first content in one of a first knowledge graph and a second knowledge graph in the preset knowledge graph, where one of the first knowledge graph and the second knowledge graph records an attribute value of at least one attribute of a first object, and the other records an attribute value of at least one attribute of a second object, and each knowledge graph takes object identification information of each recorded object and the attribute value of each attribute of the object as nodes and takes the attribute of the object as an edge to connect the nodes;

the second obtaining submodule is used for obtaining the label information of the first node, and the label information of the first node is used for indicating the object requirement of first content adaptation;

a third obtaining sub-module, configured to obtain a second node in another one of the first knowledge-graph and the second knowledge-graph based on the tag information of the first node, and use a content corresponding to the second node as a content having a specific relationship with the first content, where the tag information of the second node and the tag information of the first node indicate the same meaning.

Preferably, the second obtaining unit further includes:

a merging module, configured to connect the first node and the second node via an edge, so as to form a knowledge-graph from the first knowledge-graph and the second knowledge-graph.

Preferably, the third obtaining unit is specifically configured to combine the first content and the second content into the feedback information;

or

The application also discloses an electronic device comprising a processor and a display component, wherein the processor is provided with an intelligent conversation system which can respond to the received input information and provide feedback information;

the processor is used for obtaining input information, obtaining first content matched with the input information and second content associated with the first content, and obtaining feedback information based on the first content and the second content associated with the first content, wherein the input information is information input by a user in a human-computer conversation process through the intelligent conversation system;

the display component is used for outputting the feedback information.

The application also discloses a storage medium, wherein the storage medium is stored with computer program codes, and the computer program codes realize the data processing method when executed.

According to the technical scheme, after the input information is obtained, the first content matched with the input information and the second content associated with the first content are obtained, the feedback information is obtained based on the first content and the second content associated with the first content, and the feedback information is output, so that the feedback information meeting the expectation of the user is output based on the input information. The first content and the second content are both related to input information, and the input information can be a user problem provided by a user for a user requirement, so that the first content and the second content can be matched with the user requirement, that is, the feedback information obtained based on the first content and the second content can effectively solve the user requirement of the input information of the man-machine conversation, further, the user requirement of the man-machine conversation can be effectively solved, and the user experience is improved.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

FIG. 1 is a schematic diagram of an intelligent conversational system disclosed in embodiments of the application;

FIG. 2 is a flow chart of a data processing method disclosed in an embodiment of the present application;

FIG. 3 is a flow chart of a method for obtaining second content disclosed in an embodiment of the present application;

FIG. 4a is a schematic diagram of a default knowledge-graph as disclosed in an embodiment of the present application;

FIG. 4b is a diagram illustrating a form of feedback information disclosed in an embodiment of the present application;

FIG. 4c is a schematic illustration of another default knowledge-map disclosed in an embodiment of the present application;

fig. 4d is a schematic diagram of another form of feedback information disclosed in an embodiment of the present application;

FIG. 5 is a schematic diagram of another method for obtaining second content disclosed in the embodiments of the present application;

FIG. 6a is a schematic diagram of a default knowledge-graph as disclosed in an embodiment of the present application;

FIG. 6b is a diagram illustrating a form of feedback information disclosed in an embodiment of the present application;

FIG. 6c is a schematic illustration of another default knowledge-map disclosed in an embodiment of the present application;

fig. 6d is a schematic diagram of another form of feedback information disclosed in an embodiment of the present application;

FIG. 6e is a schematic illustration of yet another default knowledge-graph as disclosed in an embodiment of the present application;

FIG. 7 is a schematic structural diagram of a data processing apparatus according to an embodiment of the present disclosure;

FIG. 8 is a schematic diagram of another data processing apparatus according to an embodiment of the present disclosure;

fig. 9 is a schematic structural diagram of still another data processing apparatus disclosed in an embodiment of the present application;

fig. 10 is a schematic structural diagram of another data processing apparatus according to an embodiment of the present application.

Detailed Description

Currently, in the process of performing a human-computer conversation through an intelligent conversation system, feedback information output through the intelligent conversation system may have a gap from a reply expected to be received by a user, for example, the intelligent conversation system is applied to an electronic device, for example, the intelligent conversation system is installed in the electronic device in the form of an APP (application program), and when the APP is run by the electronic device, a human-computer conversation can be performed by the user through the electronic device.

As shown in fig. 1, when the user asks a user question (which may be regarded as one kind of input information) of "screen size of product a" through the intelligent conversation system, the intelligent conversation system outputs feedback information of "5.5 inches". However, the actual requirement that the user asks "the screen size of product a" is to know whether the mobile phone of the model is suitable for being held by one hand, and obviously, the feedback information output by the intelligent session system cannot meet the requirement of the user, in this case, the current man-machine session may be ended due to the difference between the feedback information and the reply expected to be received by the user, or the intelligent session system may receive the input information that "i want to know the size of the mobile phone screen and see whether the mobile phone is suitable for being held by one hand", as shown in fig. 1, the feedback information output by the intelligent session system can meet the requirement of the user through multiple interactions, but the use experience of the user is affected by the way.

Therefore, the embodiment of the application discloses a data processing method and device, which are used for outputting feedback information meeting the expected expectation of a user based on input information of the user, namely the feedback information obtained based on the first content and the second content can effectively meet the user requirement of the man-machine conversation, so that the interaction times are reduced, and the use experience of the user is improved.

In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

Referring to fig. 2, a data processing method provided by an embodiment of the present application is shown, where the data processing method is applied to an intelligent session system, and the intelligent session system is capable of responding to received input information and providing feedback information, and may include the following steps:

201: input information is obtained.

In this embodiment, the input information is information input by a user during a human-computer conversation performed by the intelligent conversation system, and the input information is information input by the user to the intelligent conversation system according to the user's own desire.

The input information may include at least one of issue information, data information, and demand information, etc. The question information is inquired for an object, for example, the value of at least one attribute of an object is inquired, the data information is the object which gives the value of at least one attribute to inquire the value of the attribute, and the requirement information is the current requirement of the user directly. The following are exemplified separately: on one hand, for electronic products, for example, corresponding to question information, the user's input may be "how large is the battery of a certain model of mobile phone? For example, the input information may be "which product has a battery of 2000mA and a memory of 64G" or "demand information," and the input information may be "demand information such as" cruising ability of large screen, "for example, the input information may be" which product has a strong cruising ability "or" how large cruising ability of product a ".

On the other hand, for the clothing product, the input of the user may be question information such as "whether the size is normal or not, whether the size is suitable for a person with a fat body shape" or not, the input of the user may be data information such as "height 180, weight 70 kg" or the like according to the data information, and the input of the user may be requirement information such as "warm keeping, static prevention" or the like according to the requirement information.

According to the contents of the above two aspects, the specific contents of the input information may be different for different types of objects, but the types of the input information include, but are not limited to, question information, data information, and requirement information, and the rest of the information and the types are not listed.

In practical applications, the intelligent conversation system can process input information through NLUPipeline (natural language understanding model), and the processing procedure may include: converting input information into Word vectors by using a Word2Vec (Word vector model), segmenting words, annotating parts of speech, analyzing sentences, understanding intentions, extracting key information and the like to finally obtain key-value label pairs consisting of processing results, so that the intelligent conversation system can obtain feedback information matched with the input information based on the key-value label pairs.

For example, in the process of performing a human-computer conversation, the input information is M, and M is processed in NLUPipeline, where M is obtained by first obtaining M { M1, M2, …, mn }, each element in M represents a word in the input information, and then obtaining a processing result O { O1, O2, …, on }, each element in O represents a pair of labels obtained after the input information is subjected to language understanding, that is, the above-mentioned key-value label pair. It should be noted that the above mentioned processes for processing the input information include, but are not limited to, the above several ways, and the rest ways are not listed here.

202: first content matching the input information and second content associated with the first content are obtained. That is, the first content is directly obtained through the input information, and the second content needs to be obtained through the first content, and the second content is indirectly related to the input information, so that the content related to the input information is expanded.

In this embodiment, a first possible manner of the first content and the second content is: one of the first content and the second content is an attribute value of a first attribute of the first object, and the other is an attribute value of a second attribute of the first object, the first attribute and the second attribute being different. Namely, the following cases are included: the first content is an attribute value of a first attribute of the first object, and the second content is an attribute value of a second attribute of the first object. Or the first content is the attribute value of the second attribute of the first object, and the second content is the attribute value of the first attribute of the first object. For example, when the input information is "what the screen size of product a is", the intelligent conversation system outputs "screen size: 5.5 inches, resolution: 2248 × 1080 ", wherein the screen size is the first content related to the first content, and the resolution is the second content related to the second content.

The second way of the first content and the second content is: one of the first content and the second content is an attribute value of a first attribute of the first object, and the other is an attribute value of a first attribute of the second object, the first object and the second object being different. Namely, the following conditions are included: the first content is an attribute value of a first attribute of the first object, and the second content is an attribute value of a first attribute of the second object. Or the first content is an attribute value of a first attribute of the second object, and the second content is an attribute value of the first attribute of the first object. For example, when the input information is "what the screen size of product a is", the intelligent conversation system outputs "the screen size is: 5.5 inches, palm fit length: 18cm "of feedback message, wherein the screen size is the first content referred to above, and the palm length is the second content referred to above.

A third way of the first content and the second content is: one of the first content and the second content is object identification information of a first object having an attribute value of a first attribute, and the other is object identification information of a second object having an attribute value of the first attribute. For example, when the input information is "5.5 inch screen cell phone", the intelligent session system outputs "cell phone with 5.5 inch screen is product a and product B", where product a is the first content mentioned above and product B is the second content mentioned above.

The above three ways are merely illustrative, and other contents may also be given in practical applications, for example, for the first way, a third content associated with the first content may also be obtained, the third content may be an attribute value of another attribute belonging to the same object as the first content, and for the third way, an attribute value of another attribute of each object may also be obtained in addition to obtaining object identification information of different objects, for example, the first content and the second content indicate an object 1 and an object 2 having an attribute value of an attribute B, and then an attribute value of an attribute C of the object 1 and/or the object 2 may also be obtained on the basis.

In this embodiment, the manner of obtaining the first content and the second content associated with the first content is, but not limited to: the method comprises the steps of firstly obtaining first content matched with user input information from a preset knowledge graph, and then obtaining content having a specific relation with the first content from the preset knowledge graph, wherein the content having a characteristic relation with the first content is second content related to the first content.

For example, according to the processing result O ═ O1, O2, …, on } obtained by processing M as user input information as described above, a fact knowledge graph (a part of the preset knowledge graph) K ═ E, R > matching the input information is obtained in the preset knowledge graph, where E ═ E1, E2, …, en }, each element in E represents an entity in the fact knowledge graph, R ═ R1, R2, …, rn }, each element in R represents a relationship between some two entities in the fact knowledge graph, and thus the first content can be obtained by E in the fact knowledge graph. Obtaining a result graph G ═ E ', R ' >, where E ' ═ E ' 1, E ' 2, …, E ' n }, each element in E ' represents an entity in the result graph, R ' ═ { R ' 1, R ' 2, …, R ' n }, and each element in R ' represents a relationship between some two entities in the result graph, so that a first content and a second content can be obtained through E ' in the result graph G, where the same entity contained in E ' of the result graph G and E of the fact knowledge graph K can be regarded as a first content, an entity in the result graph G having a relationship with the first content can be regarded as a second content, and a relationship with the first content can be represented by R ' in the result graph.

203: feedback information is obtained based on the first content and second content associated with the first content.

In this embodiment, the first content and the second content are combined into the feedback information, or the first content and the second content are modified to obtain the feedback information.

For example, without modifying the first content and the second content, the first content and the second content are directly combined into feedback information a ═ { a1, a2, …, an }, where each element in a represents a word of the feedback information, and each word is a word in the first content and the second content, for example, the first content and the second content are: 5.5 inches and 64G, the feedback information may be: 5.5 inches, 64G, making the description of the feedback information simple, but there is a problem: it is possible that after the feedback information is output, the user does not know what each part of the feedback information represents, for example, for the above-mentioned feedback information "5.5 inches, 64G", if the user does not know about the attribute of the product, the user cannot know what these two values represent with the feedback information at a glance.

Another example of a way to modify the first content and the second content is: selecting a corresponding language template according to the type of the input information, filling the first content and the second content into the language template, and correcting syntax errors in the language template in a preset rule matching mode to obtain feedback information. The type of the input information indicates the mood of the input information, for example, the input information may be any one of a statement sentence, a question sentence and a double negative sentence, and a language template for indicating the modification mode of the first content and the second content after the first content and the second content are obtained is configured in advance for the sentences of the type. For example, positions for filling the first content and the second content are reserved in the language template, so that after the first content and the second content are obtained, the first content and the second content can be filled into the corresponding positions to obtain the feedback information.

In addition, for the language template filled with the first content and the second content, the grammar error of the language template is modified through a preset rule, so that the finally obtained feedback information meets the language requirement, wherein the preset rule can be set according to the actual requirement, and the embodiment is not explained again. After obtaining the above-mentioned first and second contents "5.5 inches and 64G" in this way, the obtained feedback information may be, but is not limited to: the screen size of the product A is 5.5 inches, and the storage memory is 64G, so that the requirements of a large screen and high memory of a user can be met.

204: and outputting the feedback information.

In this embodiment, the manner of outputting the feedback information may be: the feedback information is output to the electronic device through the intelligent conversation system, and then is displayed by the electronic device, for example, the feedback information is displayed on a display area through the electronic device, and the display area can belong to the electronic device or a display area extended for the electronic device. For example, the intelligent session system outputs the feedback information to the mobile phone or the tablet device, and the feedback information is displayed through a screen of the mobile phone or the tablet device. For another example, the mobile phone or the tablet device is connected with the projector, the intelligent session system outputs the feedback information to the mobile phone or the tablet device, the mobile phone or the tablet device outputs the feedback information to the projector, and the feedback information is displayed by means of the projection function of the projector, so that the user can view the feedback information.

According to the technical scheme, after the input information is obtained, the first content matched with the input information and the second content associated with the first content are obtained, the feedback information is obtained based on the first content and the second content associated with the first content, and the feedback information is output, so that the feedback information meeting the expectation of the user is output based on the input information. The first content and the second content are both related to input information, and the input information can be a user problem provided by a user for a user requirement, so that the first content and the second content can be matched with the user requirement, that is, the feedback information obtained based on the first content and the second content can effectively solve the user requirement of the input information of the man-machine conversation, further, the user requirement of the man-machine conversation can be effectively solved, the interaction times can be reduced, and the user experience is improved.

Referring to fig. 3, for a process of obtaining a content having a specific relationship with a first content in a preset knowledge graph, a possible way of obtaining a second content disclosed in an embodiment of the present application is shown, which may include the following steps:

301: a first node of first content is obtained in a preset knowledge graph.

In this embodiment, the preset knowledge graph records an attribute value of at least one attribute of at least one object, and the preset knowledge graph takes the object identification information and the attribute value of each attribute of the object as nodes and takes the attribute of the object as an edge to connect the nodes.

That is, the preset knowledge graph may be composed of attribute values of a plurality of attributes of one object, or may be composed of attribute values of a plurality of attributes of a plurality of objects, and types of the plurality of objects may be the same or different when composed of a plurality of objects. For a preset knowledge graph, some nodes in all nodes of the preset knowledge graph are named by object identification information of an object, some nodes are named by attribute values of attributes of the object, and edges between the nodes are named by the attributes of the object.

For different types of objects, the attributes and the attribute values corresponding to the attributes are different. For example, for a mobile phone product, the attributes are the screen, system version, operating memory, storage memory, battery, and system on chip (Soc) attributes of the mobile phone product. For clothing products, the attributes are the material, size, washing mode and the like of the clothing. In this embodiment, the attributes of the remaining types of products are not all enumerated.

302: and obtaining a second node which is directly connected with the first node corresponding to the first content in the preset knowledge graph.

In this embodiment, the direct connection indicates that the first node and the second node are connected through an edge, for example, in a preset knowledge graph, if the nodes are connected by using an attribute as an edge, one of the first node and the second node is an attribute value of the attribute, and the other is object identification information.

303: and acquiring a third node which is directly connected with the second node in the preset knowledge graph, and taking the content corresponding to the third node as the content having the specific relationship with the first content, so that the second node in the preset knowledge graph can be taken as a connecting medium, and the content having the specific relationship with the first content can be acquired through the second node.

In this embodiment, when the preset knowledge graph is composed of attributes of an object, if the object includes a plurality of attributes and a third node directly connected to the second node includes attribute values of more than one attribute, the attribute values of the plurality of attributes except for the attribute corresponding to the first node can be obtained by the second node, but if the object composing the preset knowledge graph may only have one attribute, after the attribute is taken as the first node, the number of the third nodes directly connected to the object B may be 0, and the content having a specific relationship with the first content obtained at this time is empty.

When the preset knowledge graph is composed of attributes of a plurality of objects, the third node is an attribute value of the attribute of a different object. For example, when the input information is "a mobile phone with a screen size of 5.5 inches", the attribute of the first node is obtained as a screen, the attribute value is 5.5 inches, and if the mobile phone corresponding to the 5.5-inch screen has a product a and a product B, and the product a and the product B are second nodes, the third node is an attribute value of an attribute directly connected to the product a and an attribute value of an attribute directly connected to the product B.

As can be seen from the above, if the preset knowledge-graph is composed of one object, and the object includes a plurality of attributes, the content having a specific relationship with the first content obtained by the second node is: attribute values of other attributes belonging to one object with the first content, and if the preset knowledge graph is composed of a plurality of objects each including a plurality of attributes, the content having a specific relationship with the first content obtained by the second node is: the attribute values of other attributes belonging to one object with the first content, and the attribute values of other attributes belonging to a different object with the first content, wherein the other attributes refer to attributes other than the attribute corresponding to the first content.

In this embodiment, to better explain how to obtain the content having a specific relationship with the first content in the preset knowledge graph, the following description is given with reference to fig. 4a to 4 d:

example one:

the preset knowledge graph shown in fig. 4a is an example of a product a, and a preset knowledge graph is formed by attribute values of multiple attributes of the product a, where in fig. 4 a: the attribute of the node 401 is a screen, the attribute value is 5.5 inches (inch), the node 402 is an object, the object identification information is product a, the attribute of the node 4031 is a system version, the attribute value is 6.0.1, the attribute of the node 4032 is a battery, the attribute value is 2000 milliampere-hour (mAh), the attribute of the node 4033 is a storage memory, the attribute value is 64G, the attribute of the node 4034 is Soc, the attribute value is 1.8GHz, the attribute of the node 4035 is a storage memory, and the attribute value is 4 GB.

Assume that a first node 401 corresponding to a first content is acquired from a preset knowledge graph, and a second node 402 directly connected with the first node 401 is acquired, where the first node 401 and the second node 402 are directly connected by taking an attribute of the first node 401 as an edge. Obtaining a third node directly connected to the second node 402, the third node being: node 4031, node 4032, node 4033, node 4034, and node 4035. And taking the content corresponding to the third node, namely the attribute and the attribute value of the third node as the content having the specific relationship with the first content.

Referring to fig. 4b, for example, when the user inputs "mobile phone screen size of product a" information, the intelligent session system obtains attribute values of the screen size and attribute values of attributes other than the screen size from the preset knowledge map, such as the attribute values obtained in fig. 4 b: "screen size: 5.5 inch, system version: 6.0.1, battery: 2000mAh, storage memory: 64GB, running memory: 4GB, system chip information: 1.8 GHz' and the like, and then feedback information output is obtained by combining the attribute values, and the specific feedback mode can directly send the information to a user. The information can also be nested into a language template to form a complete sentence output, for example, the information is combined into a sentence "a mobile phone belongs to a large-screen mobile phone, the screen size is: 5.5 inches, system version: 6.0.1, cell size: 2000mAh, the storage memory is: 64GB, the running memory is: 4GB, system chip information is: 1.8GHz ", and outputting the sentence through the intelligent conversation system.

Example two:

the preset knowledge graph shown in fig. 4c is an example of a product a and a product B, and a preset knowledge graph is formed by attribute values of a plurality of attributes of the product a and the product B. In fig. 4c, the attribute of the node 401 is a screen, the attribute value is 5.5 inches (inch), the node 402 is an object, the object identification information is product a, the attribute of the node 4021 is a system version, the attribute value is 6.0.1, the attribute of the node 4022 is an operating memory, the attribute value is 4GB, the attribute of the node 4023 is a storage memory, the attribute value is 64GB, the node 403 is an object, the object identification information is product B, the attribute of the node 4031 is a system version, the attribute value is 6.0.1, the attribute of the node 4032 is an operating memory, the attribute value is 6GB, the attribute of the node 4033 is a storage memory, and the attribute value is 128 GB.

Assume that a first node 401 corresponding to first content is acquired in a preset knowledge graph, a second node directly connected to the first node 401 is acquired as a node 402 and a node 403, a third node directly connected to the node 402 is acquired as a node 4021, a node 4022 and a node 4023, and a third node directly connected to the node 403 is acquired as a node 4031, a node 4032 and a node 4033. And taking the content corresponding to the third node, namely the attribute and the attribute value of the third node as the content having the specific relationship with the first content.

Referring to fig. 4d, for example, when the user inputs "cell phone with 5.5 inches screen" information, the intelligent session system will send "cell phone model: the screen sizes of the product A and the product B and the product A and the product B are as follows: 5.5 inch, product A and product B system versions: 6.0.1, the operating memory of product A is: 4GB, the storage memory of the product A is: the 64GB and the running memory of the product B are as follows: the 6GB and the product B have the storage memories as follows: 128GB ", etc., and the specific feedback mode can directly output the information as feedback information. The information can also be nested into a language template to form a complete sentence for output, for example, the information is combined into a sentence "a mobile phone with a 5.5 inch mobile phone screen has a product a and a product B, and the product parameters of the product a are as follows: the system version is 6.0.1, the operating memory is 4GB, and the storage memory is 64GB, and the product parameters of the product B are as follows: the system version is 6.0.1, the run memory is 6GB and the storage memory is 128 GB. ", and outputs the sentence.

It should be noted that the contents shown in the above two examples are only for illustration.

According to the technical scheme, after the input information in the process of man-machine conversation through the intelligent conversation system is obtained, the first node of the first content is obtained in the preset knowledge graph, the second node directly connected with the first node is obtained in the preset knowledge graph, the content corresponding to the third node directly connected with the second node in the preset knowledge graph is used as the second content having a specific relation with the first content, the feedback information is obtained, and the feedback information is output. Therefore, feedback information which meets the expected expectation of the user is output based on the input information, the first content and the second content can be matched with the user requirements, namely the first content and the second content can effectively meet the user requirements of the man-machine conversation, the interaction times can be reduced, and the user use experience is improved.

Another way to obtain content having a specific relationship with the first content in the preset knowledge-graph is shown in fig. 5, which may include the following steps:

501: a first node corresponding to the first content is obtained in one of a first knowledge-graph and a second knowledge-graph in a preset knowledge-graph.

In this embodiment, one of the first knowledge graph and the second knowledge graph records an attribute value of at least one attribute of the first object, and the other records an attribute value of at least one attribute of the second object, and each knowledge graph takes the object identification information of the respective recorded object and the attribute value of each attribute of the object as nodes, and takes the attribute of the object as an edge to connect the nodes.

502: tag information of the first node is obtained.

In this embodiment, the tag information of the first node is used to indicate the object requirement of the first content adaptation. The tag information of the first node can be set in the first knowledge graph and the second knowledge graph construction process, for example, information capable of reflecting the object requirement is obtained from evaluation information used in the first knowledge graph and the second knowledge graph construction process, and the tag information is obtained based on the information capable of reflecting the object requirement, for example, the information capable of reflecting the object requirement is directly used as the tag information. For example, the information representing the object requirements is: the method is suitable for the requirement of a large screen, and the requirement suitable for the large screen can be used as label information.

503: and obtaining a second node in the other of the first knowledge graph and the second knowledge graph based on the label information of the first node, and taking the content corresponding to the second node as the content having a specific relation with the first content.

In this embodiment, the tag information of the second node and the tag information of the first node indicate the same meaning, so that the content corresponding to the second node having the same tag information as the first node is obtained, and the two-aspect content corresponding to the same object requirement is obtained. After obtaining the content having the specific relationship with the first content based on the first knowledge-graph and the second knowledge-graph, the data processing method provided by this embodiment may further include: the first node and the second node are connected by an edge to form a knowledge-graph from the first knowledge-graph and the second knowledge-graph, so that the content having a specific relationship with the first content can be obtained by using the method shown in fig. 3 when obtaining the content based on the first knowledge-graph and the second knowledge-graph again.

In this embodiment, to better explain how to obtain the content having a specific relationship with the first content in the preset knowledge graph, the following examples are used to illustrate, with specific reference to fig. 6a to 6 e:

example one: the method comprises the steps of obtaining a first node corresponding to first content from a preset first knowledge graph or a preset second knowledge graph, obtaining label information of the first node, obtaining a second node of which the label information is consistent with the label information of the first node in the other knowledge graph according to the label information of the first node, and taking content corresponding to the second node as content having a specific relation with the first content.

For example, referring to fig. 6a, in the first knowledge-graph, the node 601 is an object, the object identification information is a, the attribute of the node 602 is the screen size, the attribute value is 5.5 inches, 603 is a label, and the label information is a large screen. In the second knowledge graph, the attribute of the node 604 is the palm size, the attribute value is 18cm, 605 is a label, and the label information is a large screen. The second knowledge graph in fig. 6a is a pre-constructed knowledge graph containing user information, which includes but is not limited to: height, weight, gender, circumference, palm size, sole size, etc.

Referring to fig. 6a, assuming that the input information is "a screen size", a first node 602 corresponding to the information is obtained from the first knowledge graph, and tag information of the first node 602 is obtained, where the content corresponding to the tag 603 is tag information. Based on the label 603, label information having the same meaning as the information of the label 603 and a corresponding node are searched from the second knowledge graph, and a second node 604 and a label 605 are obtained. The content corresponding to the second node 604 is taken as the content having a specific relationship with the first content.

That is, when the user inputs the information of "screen size of a", the intelligent conversation system will compare "screen size: 5.5 inches, palm size: and (4) feeding back information such as 18 cm' to the user, wherein the information can be directly output in a specific feedback mode. The information may also be nested into a language template to form a complete sentence output, for example, in fig. 6b, the information is combined into a sentence "according to your input information, the screen size of a is: 5.5 inches, palm size: 18cm "and outputs the sentence.

Another example is: the method comprises the steps of obtaining a first node corresponding to first content from a preset first knowledge graph or a preset second knowledge graph, obtaining a label of the first node, obtaining a label of which the label information is consistent with the label information of the first node and a corresponding second node from another knowledge graph according to the label information of the label of the first node, selecting at least one third node directly connected with the second node from another knowledge graph, and taking content corresponding to the third node and/or content corresponding to a fourth node directly connected with the third node from another knowledge graph as content having a specific relation with the first content.

The method for selecting at least one third node directly connected to the second node in another knowledge-graph includes, but is not limited to: and selecting one of all nodes directly connected with the second node as a third node, or outputting object identification information corresponding to all nodes directly connected with the second node, prompting a user to select from all the object identification information, and taking the node corresponding to the selected object identification information as the third node. The following description takes an example in which the user selects the third node and takes the content corresponding to the fourth node directly connected to the third node as the content having the specific relationship with the first content:

for example, referring to fig. 6c, in the first knowledge-graph, the attribute of the node 601 is the palm size, the attribute value is 18cm, and the label information of the label 602 is a large screen. In the second knowledge graph, the label information of the label 603 is a large screen, the attribute of the node 604 is the screen size, the attribute value is 5.5 inches, the node 605 is the object, the object identification information is the product a and the node 606 are the objects, the object identification information is the product B and the node 607 are the objects, and the object identification information is the product C.

In connection with fig. 6c, assuming that the user inputs information of "18 cm in hand length", a first node 601 is obtained from the first knowledge-graph, a label 602 of the first node 601 is obtained, a label 603 corresponding to the label information of the label 602 is obtained from the second knowledge-graph according to the label 602, and a second node 604 is obtained from the second knowledge-graph according to the label 603. Acquiring all objects directly connected to the second node 604: node 605, node 606 and node 607, and output in the form of drop-down box with any model identifier in the model series in fig. 6d, and the user can select a model identifier from the drop-down box. For example, the user selects the model of the product a, that is, the selected third node is the node 605, the node directly connected to the third node 605 is the fourth node, and the content corresponding to the fourth node is the content having the specific relationship with the first content, so as to obtain the content shown in fig. 6 e.

In fig. 6e, node 605 is the object, the object identification information is product a, the attribute of node 6051 is the system version, the attribute value is 6.0.1, the attribute of node 6052 is the battery, the attribute value is 2000 milliampere-hour (mAh), the attribute of node 6053 is the storage memory, the attribute value is 64G, the attribute of node 6054 is Soc, the attribute value is 1.8GHz, the attribute of node 6055 is the storage memory, and the attribute value is 4 GB. The nodes 6051 to 6055 are set as fourth nodes, and the content corresponding to the fourth nodes is set as content having a specific relationship with the first content.

With reference to fig. 6d, for example, when the user inputs the information "18 cm in hand length", the intelligent conversation system will feed back to the user "according to the information of your hand length, the screen size of the mobile phone that you fit to is 5.5 inches, the screen size of the following mobile phone model is 5.5 inches, and you select the model that you want to know: product a, product B, and product C ", when the user selects" product a ", the intelligent session system will" mobile phone model: A. screen size: 5.5 inch, system version: 6.0.1, battery: 2000mAh, storage memory: 64GB, running memory: 4GB, system chip information: 1.8GHz, characteristics: and (3) outputting information such as large-screen mobile phones and the like, wherein the specific feedback mode can directly output the information. The information can also be nested into a language template to form a complete sentence for output, for example, the information is combined into a sentence "the model number selected by you is a, a mobile phone belongs to a large-screen mobile phone, and the screen size is: 5.5 inches, system version: 6.0.1, cell size: 2000mAh, the storage memory is: 64GB, the running memory is: 4GB, system chip information is: 1.8GHz ", and outputs the above sentence.

It should be noted that the contents mentioned in the above two examples are only for illustration.

According to the technical scheme, after the input information in the process of man-machine conversation through the intelligent conversation system is obtained, label information of a first node and a first node corresponding to first content is obtained in one of a first knowledge graph and a second knowledge graph in a preset knowledge graph, a second node is obtained in the other one of the first knowledge graph and the second knowledge graph based on the label information of the first node, the content corresponding to the second node is used as second content having a specific relation with the first content, feedback information is obtained, and the feedback information is output. Therefore, feedback information which meets the expected expectation of the user is output based on the input information, the first content and the second content can be matched with the user requirements, namely the first content and the second content can effectively meet the user requirements of the man-machine conversation, the interaction times can be reduced, and the user use experience is improved.

While, for purposes of simplicity of explanation, the foregoing method embodiments have been described as a series of acts or combination of acts, it will be appreciated by those skilled in the art that the present application is not limited by the order of acts or acts described, as some steps may occur in other orders or concurrently with other steps in accordance with the application. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required in this application.

Corresponding to the foregoing method embodiments, an embodiment of the present application discloses a data processing apparatus, which is applied to an intelligent session system, where the intelligent session system is capable of responding to received input information and providing feedback information, and a structure of the data processing apparatus is shown in fig. 7, and may include: a first obtaining unit 701, a second obtaining unit 702, a third obtaining unit 703 and an output unit 704.

A first obtaining unit 701, configured to obtain input information, where the input information is information input by a user during a human-computer conversation performed by an intelligent conversation system.

In this embodiment, the input information is information input by a user during a human-computer conversation performed by the intelligent conversation system, and the input information is information input by the user to the intelligent conversation system according to the user's own desire. For the introduction of the specific input information, please refer to the description in the above method embodiments, which is not described in detail herein.

A second obtaining unit 702, configured to obtain a first content matching the input information and a second content associated with the first content.

In this embodiment, one of the first content and the second content is an attribute value of a first attribute of the first object, and the other is an attribute value of a second attribute of the first object, and the first attribute and the second attribute are different. Or, one of the first content and the second content is an attribute value of a first attribute of the first object, and the other is an attribute value of a first attribute of the second object, and the first object and the second object are different. Or one of the first content and the second content is object identification information of a first object having an attribute value of the first attribute, and the other is object identification information of a second object having an attribute value of the first attribute. For a detailed description, please refer to the description of the above method embodiments, which will not be described in detail herein.

A third obtaining unit 703 is configured to obtain the feedback information based on the first content and the second content associated with the first content.

In this embodiment, the first content and the second content are combined into the feedback information, or the first content and the second content are modified to obtain the feedback information. For a detailed description, please refer to the description of the above method embodiments, which will not be described in detail herein.

An output unit 704 for outputting the feedback information.

According to the technical scheme, after the input information is obtained, the first content matched with the input information and the second content associated with the first content are obtained. And obtaining feedback information based on the first content and second content associated with the first content, and outputting the feedback information. Thereby, feedback information which is expected to be expected by the user is output based on the input information of the user. The first content and the second content are related to user input information, namely the first content and the second content can effectively meet the user requirements of the input information of the man-machine conversation, so that the problems brought forward by a user can be effectively solved, and the user experience is improved.

In this embodiment, one way for the second obtaining unit 702 to obtain the second content is: and obtaining first content matched with the input information from a preset knowledge graph. And obtaining the content having a specific relation with the first content in the preset knowledge graph, wherein the content having a characteristic relation with the first content is the second content associated with the first content. Two corresponding alternative structures of the second obtaining unit 702 are shown in fig. 8, where fig. 8 shows a block diagram of another data processing apparatus disclosed in the embodiment of the present application, and in fig. 8, the second obtaining unit 702 includes: a first acquisition module 7021 and a second acquisition module 7022.

A first obtaining module 7021, configured to obtain, from a preset knowledge graph, a first content matched with the input information.

A second obtaining module 7022, configured to obtain, in the preset knowledge graph, content having a specific relationship with the first content, where the content having a characteristic relationship with the first content is a second content associated with the first content.

In an embodiment, the second obtaining module 7022 is specifically configured to obtain a first node corresponding to a first content in a preset knowledge graph, where the preset knowledge graph records an attribute value of at least one attribute of at least one object, and the preset knowledge graph takes object identification information and the attribute value of each attribute of the object as nodes and takes the attribute of the object as an edge connection node. And a second node used for obtaining a first node direct connection corresponding to the first content in the preset knowledge graph, wherein the direct connection indicates that the first node and the second node are connected through an edge. And the third node is used for obtaining the third node which is directly connected with the second node in the preset knowledge graph, and the content corresponding to the third node is used as the content having a specific relation with the first content. For a detailed description, please refer to the description of the above method embodiments, which will not be described in detail herein.

In another embodiment, the second obtaining module 7022 is specifically configured to obtain a first node corresponding to the first content in one of a first knowledge graph and a second knowledge graph in the preset knowledge graph, where one of the first knowledge graph and the second knowledge graph records an attribute value of at least one attribute of the first object, and the other records an attribute value of at least one attribute of the second object, and each knowledge graph takes the object identification information of the respective recorded object and the attribute value of each attribute of the object as nodes, and connects the nodes with the attributes of the object as edges. And the tag information of the first node is used for indicating the object requirement of the first content adaptation, and the tag information of the first node is used for obtaining a second node in the other one of the first knowledge graph and the second knowledge graph based on the tag information of the first node, taking the content corresponding to the second node as the content having a specific relation with the first content, and the tag information of the second node and the tag information of the first node indicate the same meaning. For a detailed description, please refer to the description of the above method embodiments, which will not be described in detail herein.

For another implementation, in the data processing apparatus disclosed in this embodiment of the present application, on the basis of fig. 8, the second obtaining unit 702 further includes a merging module 7023, as shown in fig. 9, where the merging module 7023 is configured to connect the first node and the second node by an edge, so as to configure the first knowledge-graph and the second knowledge-graph into a knowledge-graph.

To better explain the data processing device related to the above embodiments, this embodiment shows a possible architecture diagram of the data processing device in practical application, as shown in fig. 10, which may include: an input module 1001, a language understanding module 1002, a knowledge graph using module 1003, a knowledge relationship reasoning module 1004, a language generating module 1005, and an output module 1006. The input module 1001 and the language understanding module 1002 correspond to the first obtaining unit 701, the knowledge graph using module 1003 and the knowledge relationship reasoning module 1004 correspond to the second obtaining unit 702, the language generating module 1005 corresponds to the third obtaining unit 703, and the output module 1006 corresponds to the output unit 704.

In practical application, the input module 1001 inputs a question M (input information) into the intelligent conversation system, and the language understanding module 1002 extracts feature information from the question M and labels important words, such as product information and question types, to obtain an identification result O.

The knowledge graph using module 1003 searches entities and relationships between the entities related to the problem M, for example, information such as weight of a mobile phone product, related parameters of screen size, gender, height, weight, and palm size of the user, from the constructed fact knowledge data graph and the user portrait data graph according to the recognition result O. Based on the entities E and the relationships R between the entities obtained by the knowledge graph using module 1003, the knowledge relationship reasoning module 1004 performs reasoning to obtain a result graph G for generating an answer. The fact knowledge data picture and the user portrait data map are two knowledge maps in the preset knowledge map.

The language generation module 1005 selects a suitable language template according to the result graph G, fills the content in the result graph G into the template to generate a structured text answer a (i.e. feedback information), and outputs the text answer a through the output module 1006.

For the specific execution content of the above modules, please refer to the above method embodiments, which will not be described again.

In addition, the embodiment of the application further discloses electronic equipment which comprises a processor and a display component, wherein the processor is provided with an intelligent conversation system, and the intelligent conversation system can respond to the received input information and provide feedback information.

And the processor is used for obtaining input information, obtaining first content matched with the input information and second content associated with the first content, and obtaining feedback information based on the first content and the second content associated with the first content, wherein the input information is information input by a user in a man-machine conversation process through the intelligent conversation system. For the detailed description of the functions in the processor, reference is made to the above method embodiments, which are not described again.

And the display component is used for outputting the feedback information.

The embodiment of the application also discloses a storage medium, wherein the storage medium is stored with computer program codes, and the data processing method is realized when the computer program codes are executed.

It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the device-like embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.

Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.

The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

The foregoing is only a preferred embodiment of the present application and it should be noted that those skilled in the art can make several improvements and modifications without departing from the principle of the present application, and these improvements and modifications should also be considered as the protection scope of the present application.

Claims

1. A data processing method is applied to an intelligent conversation system, the intelligent conversation system can respond to received input information and provide feedback information, and the method comprises the following steps:

obtaining first content matched with the input information and second content associated with the first content, wherein the first content is an attribute value at least matched with the input information in a first object, the second content is an attribute value of an object except the attribute value of the first object, and the attribute values of other objects are matched with the first content;

and outputting the feedback information.

2. The method of claim 1, wherein one of the first content and the second content is an attribute value of a first attribute of a first object and the other is an attribute value of a first attribute of a second object, the first object and the second object being different.

3. The method of claim 1, the obtaining first content matching the input information and second content associated with the first content comprising:

4. The method of claim 3, the obtaining content in the preset knowledge-graph having a particular relationship with the first content comprising:

5. The method of claim 3, the obtaining content in the preset knowledge-graph having a particular relationship with the first content comprising:

6. The method of claim 5, further comprising: connecting the first node and the second node by an edge to construct the first knowledge-graph and the second knowledge-graph into a knowledge-graph.

7. The method of claim 1, the obtaining feedback information based on the first content and second content associated with the first content comprising:

or

8. A data processing apparatus for use in an intelligent conversational system, the intelligent conversational system being capable of responding to received input information and providing feedback information, the apparatus comprising:

a second obtaining unit configured to obtain a first content that matches the input information and a second content that is associated with the first content, the first content being an attribute value that matches at least the input information in a first object, the second content being an attribute value of an object other than the attribute value of the first object, and the attribute values of the other objects matching the first content;

and the output unit is used for outputting the feedback information.

9. The apparatus of claim 8, wherein one of the first content and the second content is an attribute value of a first attribute of a first object and the other is an attribute value of a first attribute of a second object, the first object and the second object being different.

10. An electronic device comprising a processor and a display component, the processor having an intelligent dialog system capable of responding to received input information and providing feedback information;

the processor is configured to obtain input information, obtain a first content that matches the input information and a second content that is associated with the first content, where the first content is an attribute value that at least matches the input information in a first object, the second content is an attribute value of an object other than the attribute value of the first object, and the attribute values of the other objects match the first content, and obtain feedback information based on the first content and the second content associated with the first content, and the input information is information input by a user during a human-computer conversation process performed by the intelligent conversation system;

the display component is used for outputting the feedback information.