CN112328741A

CN112328741A - Intelligent association reply method and device based on artificial intelligence and computer equipment

Info

Publication number: CN112328741A
Application number: CN202011209023.5A
Authority: CN
Inventors: 高毅; 袁振东
Original assignee: Ping An Technology Shenzhen Co Ltd
Current assignee: Ping An Technology Shenzhen Co Ltd
Priority date: 2020-11-03
Filing date: 2020-11-03
Publication date: 2021-02-05
Anticipated expiration: 2040-11-03
Also published as: WO2022095357A1; CN112328741B

Abstract

The invention discloses an intelligent association reply method, an intelligent association reply device and computer equipment based on artificial intelligence. The method comprises the following steps: the method comprises the steps of carrying out structural processing on mined data information according to structural processing rules to generate an associated knowledge network and adding the associated knowledge network into a knowledge graph, obtaining target subject nodes matched with received question information and historical response information associated with the question information from the knowledge graph, if the target subject nodes and the question information meet preset response conditions, obtaining data of the target subject nodes as first response information and feeding the first response information back to a user terminal, and if the preset response conditions are not met, generating node question information or obtaining bottom response information and feeding the bottom response information back to the user terminal. The invention is based on an intelligent decision technology, belongs to the field of artificial intelligence, acquires target subject nodes matched in a knowledge graph according to question information and historical response information associated with the question information, and performs intelligent associated response according to data of the target subject nodes, so that the accuracy of response to the question information can be improved.

Description

Intelligent association reply method and device based on artificial intelligence and computer equipment

Technical Field

The invention relates to the technical field of artificial intelligence, belongs to an application scene of acquiring reply information for intelligent reply in a smart city, and particularly relates to an intelligent correlation reply method, an intelligent correlation reply device and computer equipment based on artificial intelligence.

Background

With the rapid development of artificial intelligence, an enterprise can construct an intelligent interactive processing system based on artificial intelligence, for example, an intelligent question-answering system can be constructed to provide services for customers, and the question information sent by the customers is received and responded in a targeted manner through the full-time intelligent customer question-answering system, so that more convenient and rapid answering services are provided for the customers. However, the existing intelligent question-answering system only obtains and feeds back response information according to question information in the single-round question-answering process, and there is no logical relevance between multiple rounds of question-answering, however, in the practical application process, some complicated questions are difficult to obtain corresponding response information through single-round question-answering, and the intelligent question-answering system in the prior art method is difficult to correctly answer the complicated question information, so that the accuracy rate of answering the complicated questions is low. Therefore, the prior art method has the problem that the complex questions cannot be accurately answered.

Disclosure of Invention

The embodiment of the invention provides an intelligent association response method, an intelligent association response device, computer equipment and a storage medium based on artificial intelligence, and aims to solve the problem that a complex question cannot be accurately responded by the prior art.

In a first aspect, an embodiment of the present invention provides an intelligent association response method based on artificial intelligence, which includes:

if the input mining data information is received, carrying out structural processing on the mining data information according to a preset structural processing rule to generate an associated knowledge network, and adding the associated knowledge network into a prestored knowledge map;

receiving question information from the user terminal, and acquiring associated main body nodes matched with the question information and pre-stored historical question and answer information from the knowledge graph;

acquiring a target subject node matched with the question information from the associated subject node according to a preset matching model;

judging whether the target subject node and the question information meet a preset reply condition or not;

if the target subject node and the question information meet a preset reply condition, acquiring subject node data of the target subject node as first reply information and feeding back the first reply information to the user terminal;

if the target subject node and the question information do not meet a preset reply condition, judging whether the target subject node comprises a child node;

if the target main body node comprises child nodes, generating node question information according to the child nodes of the target main body node, and sending the node question information to the user terminal so as to obtain reply information fed back by the user terminal according to the node question information;

and if the target main body node does not comprise a child node, acquiring basic information corresponding to the target main body node from a pre-stored basic information database as bibliography reply information and feeding the bibliography reply information back to the user terminal.

In a second aspect, an embodiment of the present invention provides an intelligent association reply apparatus based on artificial intelligence, which includes:

the mining data information structuralization processing unit is used for structuralizing the mining data information according to a preset structuralization processing rule to generate an associated knowledge network and adding the associated knowledge network into a prestored knowledge map if the inputted mining data information is received;

the associated main body node acquisition unit is used for receiving the question information from the user terminal and acquiring the associated main body node matched with the question information and the pre-stored historical question and answer information from the knowledge graph;

a target subject node obtaining unit, configured to obtain a target subject node matched with the question information from the associated subject node according to a preset matching model;

the judging unit is used for judging whether the target subject node and the question information meet a preset reply condition or not;

the first reply information feedback unit is used for acquiring the main body node data of the target main body node as first reply information and feeding the first reply information back to the user terminal if the target main body node and the question information meet preset reply conditions;

the child node judging unit is used for judging whether the target main body node comprises a child node or not if the target main body node and the question information do not meet a preset reply condition;

a node question information sending unit, configured to generate node question information according to a child node of the target subject node and send the node question information to the user terminal if the target subject node includes the child node, so as to obtain reply information fed back by the user terminal according to the node question information;

and the pocket bottom reply information feedback unit is used for acquiring basic information corresponding to the target main body node from a pre-stored basic information database as pocket bottom reply information and feeding the pocket bottom reply information back to the user terminal if the target main body node does not comprise a child node.

In a third aspect, an embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor, when executing the computer program, implements the artificial intelligence based intelligent association reply method according to the first aspect.

In a fourth aspect, the embodiment of the present invention further provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and the computer program, when executed by a processor, causes the processor to execute the artificial intelligence based intelligent relevance responding method according to the first aspect.

The embodiment of the invention provides an intelligent association reply method and device based on artificial intelligence, computer equipment and a storage medium. The method comprises the steps of carrying out structural processing on mined data information according to structural processing rules to generate an associated knowledge network and adding the associated knowledge network into a knowledge graph, obtaining target subject nodes matched with received question information and historical response information associated with the question information from the knowledge graph, if the target subject nodes and the question information meet preset response conditions, obtaining data of the target subject nodes as first response information and feeding the first response information back to a user terminal, and if the preset response conditions are not met, generating node question information or obtaining bottom response information and feeding the bottom response information back to the user terminal. By the method, the matched target subject node in the knowledge graph can be obtained based on the question information and the historical response information associated with the question information, intelligent association response is carried out based on the data of the target subject node, and the accuracy of response to the complex question information can be improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, 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 flow chart of an artificial intelligence-based intelligent association response method according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of an application scenario of an artificial intelligence-based intelligent relevance response method according to an embodiment of the present invention;

FIG. 3 is another schematic flow chart of an artificial intelligence based intelligent relevance responding method according to an embodiment of the present invention;

FIG. 4 is a sub-flowchart of an artificial intelligence based intelligent association response method according to an embodiment of the present invention;

FIG. 5 is a schematic view of another sub-flow chart of an artificial intelligence based intelligent relevance responding method according to an embodiment of the present invention;

FIG. 6 is a schematic view of another sub-flow chart of an artificial intelligence based intelligent relevance responding method according to an embodiment of the present invention;

FIG. 7 is a schematic view of another sub-flow chart of an artificial intelligence based intelligent relevance responding method according to an embodiment of the present invention;

FIG. 8 is a schematic flow chart illustrating an artificial intelligence based intelligent relevance responding method according to an embodiment of the present invention;

FIG. 9 is a schematic block diagram of an intelligent relevance responding apparatus based on artificial intelligence provided by an embodiment of the present invention;

FIG. 10 is a schematic block diagram of a computer device provided by an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. 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 invention.

It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

It is also to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the specification of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

It should be further understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

Referring to fig. 1 and fig. 2, fig. 1 is a schematic flowchart of an artificial intelligence based intelligent association response method according to an embodiment of the present invention, and fig. 2 is a schematic application scenario diagram of the artificial intelligence based intelligent association response method according to the embodiment of the present invention; the intelligent association reply method based on artificial intelligence is applied to a management server 10, the method is executed through application software installed in the management server 10, the management server 10 is connected with at least one user terminal 20 through a network to realize data information transmission, the management server 10 is a server end used for executing the intelligent association reply method based on artificial intelligence to acquire reply information for intelligent reply, the management server can be an enterprise server, a user of the management server is an administrator of an enterprise, and the user terminal 20 is a terminal device such as a desktop computer, a notebook computer, a tablet computer or a mobile phone which is connected with the management server 10 through a network to acquire the reply information. Fig. 2 only illustrates the management server 10 performing information transmission with one user terminal 20, and in practical applications, the management server 10 may also establish communication connections with multiple user terminals 20 at the same time to implement data information transmission. As shown in fig. 1, the method includes steps S110 to S180.

S110, if the input mining data information is received, carrying out structuring processing on the mining data information according to preset structuring processing rules to generate an associated knowledge network, and adding the associated knowledge network to a pre-stored knowledge graph.

And carrying out structural processing on the mining data information according to a preset structural processing rule to generate an associated knowledge network, and adding the associated knowledge network into a pre-stored knowledge map. The structured processing rule comprises main body keyword information and main body structure relationship information. The mining data information can be subjected to structural processing according to structural processing rules, an associated knowledge network is generated and then added to the knowledge graph, wherein the structural processing rules are concrete rules for carrying out structural splitting on the mining data information to generate the associated knowledge network which is associated with each other, the associated knowledge network records the data in a tree structure mode and can reflect the association relation among the data, and the knowledge graph is an intelligent graph which is prestored in the management server and used for recording the associated knowledge network.

Specifically, in an embodiment, as shown in fig. 3, step S110 further includes steps S110a and S110b before step S110, and the mining data information may be obtained through steps S110a and S110 b.

S110a, receiving mining configuration information input by an administrator, and configuring a pre-stored data mining program according to the mining configuration information to obtain a configured data mining program.

The mining configuration information comprises mining keyword information, time information and a mining address set. The mining configuration information may be configuration information input by an administrator of the management server and used for performing parameter configuration on a data mining program, the data mining program is a computer program prestored in the management server, and designated data information corresponding to the mining configuration information may be mined according to the data mining program completing the parameter configuration.

The specific steps of configuring the parameter values in the data mining program may include: configuring key word parameters in the data mining program according to the mining key word information; configuring time parameters in the data mining program according to the time information; and configuring the mining address parameters in the data mining program according to the mining address set.

Specifically, the mining keyword information includes a plurality of mining keywords, and the data information matched with the mining keywords can be acquired through the mining keywords, for example, the mining keyword information may include "insurance clauses" and "business handling rules". The time information may be a time period for mining the data information, and to ensure timeliness of the mined data information, the storage time of the data may be limited by the time information to obtain the data information whose storage time matches the time information, for example, the time information may be 2017-2020. The mining address set includes a plurality of mining addresses, the mining addresses may be storage addresses of internet address information or files, the mining address set may be used to limit a data source range for data information mining, that is, the data information obtained by mining all includes a data source range limited by the mining address set, the mining addresses may all be represented in a URL (Uniform Resource Locator) format, for example, the internet address information may be internet protocol addresses corresponding to websites such as encyclopedic information websites, and the storage addresses of the files may be stored in specific storage locations of the files in the management server, for example, the files may be internal rule systems of an enterprise, contract files of an enterprise, and the like. Specifically, the parameters that can be configured in the data mining program include a keyword parameter, a time parameter, and a mining address parameter, and the configured data mining program can be obtained by configuring the keyword parameter according to the mining keyword information, configuring the time parameter according to the time information, and configuring the mining address parameter according to the mining address set.

S110b, executing the configured data mining program to mine the information of the mined data.

Corresponding mining data information can be obtained by executing a data mining program for completing parameter configuration, and the obtained mining data information comprises a plurality of pieces of data recorded in a text form. Specifically, character information corresponding to the mining address parameters is obtained according to the mining address parameters configured in the data mining program, if the mining address is internet address information, character information in a webpage corresponding to the internet address information is obtained, and if the mining address is a storage address of a file, character information in the file corresponding to the storage address of the file is obtained; and screening the obtained text information according to the keyword parameters and the time parameters configured in the data mining program, and acquiring the text information meeting the keyword parameters and the time parameters as mining data information matched with the mining configuration information.

In an embodiment, as shown in fig. 4, step S110 includes sub-steps S111, S112 and S113.

And S111, acquiring a main body corresponding to each piece of data in the mining data information according to the main body keyword information.

And acquiring a main body corresponding to each piece of data in the mining data information according to the main body keyword information. The main body keyword information comprises a plurality of main body keywords, and the main body keywords can be matched with data in the mining data information to obtain a main body matched with the main body keywords in the data. For example, the subject keyword may be a policyholder equity, a policy loan, or the like. If the data contains a certain main body keyword, the main body keyword is taken as a main body corresponding to the data, and the content contained in the data is description information for explaining one main body; when data includes a plurality of subject keywords, subjects corresponding to the plurality of subject keywords are all regarded as subjects of the data, and the content included in the data is information that defines the relationship between the plurality of subjects.

And S112, performing structural association on the data of each main body according to the main body structural relationship information to generate an associated knowledge network containing all the main bodies.

The main body structure relationship information comprises the association relationship among a plurality of main bodies and the structure relationship information of the classification relationship, the classification relationship can be a classification result obtained after classifying the main bodies based on an application scene, the data of each main body can be structurally associated according to the main body structure relationship information to generate a corresponding association knowledge network, specifically, the data only containing a single main body is obtained, the main body nodes corresponding to each main body are generated according to the association relationship among the main bodies in the main body structure relationship, the associated main bodies are connected in a connecting line mode, the data corresponding to each main body are stored in the main body nodes of the main body, so that the data containing the single main body can be structurally associated according to the association relationship of the main bodies to obtain the basic knowledge network containing a plurality of main body nodes; and acquiring data containing a plurality of main bodies, taking the data containing the plurality of main bodies as limited information of the incidence relation, and further limiting the incidence relation between main body nodes in the basic knowledge network to obtain the incidence knowledge network. For example, a certain data containing two principals a and B is "B of a means XXX", and a principal node corresponding to the B principal is a child node of the principal node corresponding to the a principal, a connection line for indicating a dependency relationship may be added between the two corresponding principal nodes according to the association relationship of the two principals, and the data is stored in the principal node corresponding to the principal B, and is limitation information for further limiting the B principal on the basis of the a principal.

S113, adding the associated knowledge network into the knowledge graph for storage.

The knowledge graph is an intelligent graph which is pre-stored in the management server and used for recording the associated knowledge network, other associated knowledge networks may be stored in the knowledge graph, and the obtained associated knowledge network can be added into the knowledge graph to be stored so as to supplement and increase the content in the original associated knowledge network.

And S120, receiving the question information from the user terminal, and acquiring the associated main body node matched with the question information and the pre-stored historical question and answer information from the knowledge graph.

The user can send the question information to the management server through the user terminal based on own requirements, the management server can obtain reply information corresponding to the question information to reply, the specific management server also stores historical question-answer information corresponding to the question information, the historical question-answer information is information obtained by recording the completed question-answer information between the user of the client and the management server before sending the question information, the historical question-answer information and the associated content associated with the question information can be combined to obtain a main body node matched with the question information and the associated content in a knowledge graph as an associated main body node, and the number of the associated main body nodes can be one or more.

In an embodiment, as shown in fig. 5, step S120 includes sub-steps S121 and S122.

S121, acquiring multiple rounds of historical question information which is the same as the application scene information of the question information from the historical question and answer information as associated question information; and S122, acquiring the associated main body node of the main body node matched with the associated question information in the knowledge graph.

Specifically, the question information is question information in the current round of question answering, multiple rounds of historical question information before the current round and with application scene information identical to the question information can be obtained, and the multiple rounds of historical question information with the application scene information identical to the question information is used as associated question information. Since each piece of questioning information in the historical questioning and answering information is matched with one main body node in the knowledge graph, the questioning information of the current round is most likely to be matched with other main body nodes associated with the main body node of the historical questioning and answering information, and all the main body nodes associated with the main body node associated with the questioning information in the knowledge graph can be obtained to serve as associated main body nodes.

S130, acquiring a target subject node matched with the question information from the associated subject nodes according to a preset matching model.

The target subject nodes matched with the question information in the obtained associated subject nodes can be obtained through the matching model, the associated question information matched with the question information needs to be used when the target subject nodes are obtained, and only one target subject node is provided. More specifically, the matching model comprises a vocabulary processing rule, a neural network and a similarity calculation formula. The life insurance can combine the question information and the associated question information to obtain combined information, the vocabulary processing information is a specific rule for converting the vocabulary contained in the combined information or the vocabulary contained in the node data of the main body node, the corresponding characteristic vector can be obtained after the conversion processing, the neural network is a neural network model which is constructed based on the long-short term memory network and is used for processing and analyzing the characteristic vector, the neural network is used for processing the characteristic vector to obtain the output information of the memory network, and the similarity calculation formula is a calculation formula for calculating the output information of the two groups of memory networks to obtain the similarity.

In an embodiment, as shown in fig. 6, step S130 includes sub-steps S131, S132, S133, S134, S135 and S136.

S131, combining the question information and the associated question information to obtain combined information; and S132, converting the combined information according to a preset vocabulary processing rule to obtain a corresponding question feature vector.

And processing the combined information according to the vocabulary processing rule to obtain an N multiplied by M characteristic vector corresponding to the combined information, wherein M is word vector dimension information, and M and N are positive integers larger than zero. The vocabulary processing rules comprise vocabulary screening rules, a vocabulary vector table and vocabulary quantity information, wherein the vocabulary quantity information is used for intercepting N vocabularies, namely the vocabulary quantity information is N, and the vocabulary vector table can convert any character into a vector with dimension of 1 multiplied by M. The method comprises the following specific steps: screening the combined information according to the vocabulary screening rule to obtain screened vocabulary information; standardizing the screened vocabulary information according to the vocabulary quantity information to obtain standard vocabulary information; and obtaining a question feature vector corresponding to the standard vocabulary information according to the vocabulary vector table.

The vocabulary screening rule is rule information for screening the vocabulary in the combined record information, specifically, the vocabulary screening rule can screen out the vocabulary with insusceptibility in the combined information, and the vocabulary contained in the obtained screening vocabulary information is the vocabulary with practical significance. For example, the words to be screened may be "o", "di", "me", etc. in the word screening rule. If the number of words contained in the filtered word information is a non-fixed value, the filtered word information needs to be standardized according to the word number information, so as to obtain N words corresponding to the word number information as standard word information. Specifically, if the number of vocabularies contained in the screening vocabulary information exceeds N, capturing vocabularies with the same number in the screening vocabulary information as standard vocabulary information according to N; if the vocabulary data contained in the screening vocabulary information is less than N, using a null character (represented by □) to fill up the missing vocabulary in the screening vocabulary information to obtain standard vocabulary information; and if the number of the vocabularies contained in the screening vocabulary information is just N, performing subsequent processing by using the screening vocabulary information as standard vocabulary information. The vocabulary vector table contains a 1 × M-dimensional vector corresponding to each vocabulary, and the 1 × M-dimensional vector can be used for quantifying the characteristics of a single vocabulary. According to the words contained in the standard word information, a 1 xM-dimensional vector corresponding to each word can be obtained from the word vector table, and the 1 xM-dimensional vectors corresponding to the N words are combined to obtain an NxM-dimensional vector, namely the obtained question feature vector.

And S133, converting the node data of the associated main body node according to the vocabulary processing rule to obtain a corresponding node data feature vector.

The process of converting the node data of the associated body node is the same as the process of converting the combined information, and is not described herein again.

And S134, respectively calculating the question feature vector and the node data feature vector according to the neural network to obtain corresponding question output information and node data output information.

And calculating the question feature vector according to the neural network to obtain question output information, and calculating the data feature vector of each node to obtain a plurality of corresponding node data output information. Wherein, the long-term and short-term memory network comprises a plurality of cells, and the number of the cells is equal to the vocabulary number information N. The questioning feature vectors or the node data feature vectors are N vectors with dimensions of 1 xM, and the N vectors with dimensions of 1 xM are respectively and correspondingly input into N cells of the long-term and short-term memory network. Specifically, the step of calculating the memory network output information of a certain feature vector is divided into five steps, namely calculating the forgetting gate output information: f. of^(t)＝σ(W_f×h^(t_1)+U_f×X^(t)+b_f) Wherein f is^(t)F is more than or equal to 0 to forget the door parameter value^(t)Less than or equal to 1; σ is the sign of the activation function calculation, and can be expressed specifically as

Then W will be_f×h^(t_1)+U_f×X^(t)+b_fThe calculation result of (a) is used as x input activation function sigma to calculate f^(t)；W_f、U_fAnd b_fAll are the parameter values of the formula in the cell; h is^(t_1)Output gate information for the previous cell; x^(t)Inputting a 1 xM-dimensional vector of the current cell into the feature vector, if the current cell is the first cell in the long-short term memory network, then h^(t ^_1)Is zero. Calculating input gate information: i.e. i^(t)＝σ(W_i×h^(t_1)+U_i×X^(t)+b_i)；a^(t)＝tanh(W_a×h^(t-1)+U_a×X^(t)+b_a) Wherein i^(t)Is input intoValue of gate parameter, 0 ≤ i^(t)≤1；W_i、U_i、b_i、W_a、U_aAnd b_aAre all the values of the formula in the subject cell, a^(t)For the calculated input gate vector value, a^(t)Is a vector of dimension 1 × M. Updating cell memory information: c^(t)＝C^(t_1)⊙f^(t)+i^(t)⊙a^(t)C is the cell memory information accumulated during each calculation, C^(t)Cell memory information output for the current cell, C^(t_1)Cell memory information output for the previous cell, which is a vector operator, C^(t_1)⊙f^(t)Is calculated by dividing the vector C^(t_1)Each dimension value in the above-mentioned two dimensional values is respectively equal to f^(t)Multiplication of the calculated vector dimension with the vector C^(t_1)Are the same. Fourthly, calculating output gate information: o^(t)＝σ(W_o×h^(t_1)+U_o×X^(t)+b_o)；h^(t)＝o^(t)⊙tanh(C^(t))，o^(t)Is the output gate parameter value, 0 is less than or equal to o^(t)≤1；W_o、U_oAnd b_oAre all the values of the formula in the subject cell, h^(t)Output gate information of the subject cell, h^(t)Is a vector of dimension 1 × M. Calculating the output information of the current cell: y is^(t)＝σ(V×h^(t)+ c), V and c are the values of the formula in this cell. And calculating each cell to obtain output information, and synthesizing the output information of the N cells to obtain the memory network output information of the characteristic vector, wherein the memory network output information of the characteristic vector is a 1 multiplied by N-dimensional vector.

S135, calculating the similarity between the question output information and each node data output information according to the similarity calculation formula; s136, obtaining the associated main body node of the node data characteristic vector with the highest similarity as the target main body node.

The similarity between the question output information and the data output information of each node can be calculated according to a similarity calculation formula, and the similarity with the maximum similarity is obtained from the similarity calculation formulaThe higher one of the associated subject nodes serves as the corresponding target subject node. Specifically, the similarity calculation formula may be

Wherein F ═ F₁,f₂…f_N) To output information for questioning, R ═ R (R)₁,r₂…r_N) And outputting information for data of a certain node.

S140, judging whether the target subject node and the question information meet preset reply conditions.

After the target subject node is obtained, whether the target subject node and the question information meet a preset reply condition or not needs to be judged, if the preset reply condition is met, node data of the target subject node is directly obtained to reply, and if the preset reply condition is not met, other reply modes are adopted to reply. Wherein the preset reply condition comprises a similarity threshold.

In an embodiment, as shown in fig. 7, step S140 includes sub-steps S141, S142 and S143.

S141, judging whether the similarity between the main body node data of the target main body node and the combined information is greater than a similarity threshold value in the preset reply condition or not; and S142, if the similarity between the main node data and the combined information is greater than the similarity threshold, judging whether the question information is matched with node limiting information corresponding to the main node data.

Firstly, whether a similarity threshold value between node data of a target main node and combined information meets the requirement of the similarity threshold value is judged, if yes, node limiting information corresponding to the target main node can be obtained, the node limiting information is specific information for limiting the node data contained in the target main node, the node limiting information can be composed of a node name of the main node and a node name of a father node with an affiliation with the main node, and then the node limiting information can contain a plurality of limiting keywords. Judging whether the limited keywords of the node limited information are all contained in the question information, and if the limited keywords are all contained in the question information, judging that the obtained question information is matched with the node limited information; otherwise, judging that the questioning information is not matched with the node limiting information.

For example, the node restriction information of the target subject node corresponding to the subject B is "a + B", and if the question information includes "a" and "B", the restriction keyword in the node restriction information matches the question information; if the questioning information only contains 'B', the node restriction information is not matched with the questioning information.

S143, if the question information is matched with the node limiting information of the main body node, the target main body node and the question information are judged to meet the preset reply condition. If the similarity between the subject node data and the combined information is not greater than the similarity threshold, or the question information is not matched with the node restriction information of the subject node, it may be determined that the target subject node and the question information do not satisfy the preset reply condition.

S150, if the target subject node and the question information meet the preset reply condition, acquiring the subject node data of the target subject node as first reply information and feeding the first reply information back to the user terminal.

And if the target main body node and the question information meet the preset reply condition, acquiring corresponding main body node data as first reply information and feeding the first reply information back to the user terminal to complete the question-answering process.

And S160, if the target subject node and the question information do not meet the preset reply condition, judging whether the target subject node comprises a child node.

And judging whether the target subject node contains child nodes in the knowledge graph, namely judging whether the subject nodes related to the target subject node have subject nodes subordinate to the target subject node, and obtaining a corresponding judgment result.

S170, if the target main body node comprises the child nodes, generating node question information according to the child nodes of the target main body node, and sending the node question information to the user terminal so as to obtain reply information fed back by the user terminal according to the node question information.

If the target main body node comprises the child nodes, node question information can be generated according to the child nodes of the target main body node to obtain response information fed back by the user, more detailed information related to the user can be obtained based on the response information, and more accurate response information related to the user in the knowledge graph is obtained according to the response information and the question information. The specific steps may include: and acquiring a main body name corresponding to the child node of the target main body node, generating node question information based on the main body name and sending the node question information to the user terminal.

For example, the target subject node may include a sub-node corresponding to a subject name "C", and a node questioning information "do you want to know C? "; the target subject node correspondingly comprises a plurality of sub-nodes, and the subject names corresponding to the sub-nodes are "D", "E" and "F" in sequence, so that the node questioning information "do you want to know D, E or F? "

In one embodiment, as shown in fig. 8, step S1701 is further included after step S170.

And S1701, acquiring second reply information matched with the question information and the reply information in the child nodes of the target subject node and feeding back the second reply information to the user terminal.

After the reply information is obtained, second reply information matched with the question information and the reply information in the child nodes of the target main body node can be obtained and fed back to the user terminal. Node limit information corresponding to the child node of the target subject node can be acquired, the question information and the reply information are combined, and whether the question information and the reply information are matched with the node limit information of the child node is judged again in the same mode. Specifically, the question information is the question information in the previous round of question answering process, the reply information is the information replied by the user in the current round of question answering process, the question information and the reply information can be combined, whether the node limit information of the child node is contained in the combination of the question information and the reply information is judged according to the previous process, if yes, the question information and the reply information are judged to be matched with the node limit information of the child node, and the node data of the child node is obtained to serve as second reply information; otherwise, the obtained questioning information and the reply information are not matched with the node limiting information, and corresponding prompt information can be fed back to the user terminal.

And S180, if the target main body node does not comprise a child node, acquiring basic information corresponding to the target main body node from a pre-stored basic information database as pocket bottom reply information and feeding the pocket bottom reply information back to the user terminal.

The basic information database is a database which is prestored in the management server and used for storing basic information of the main bodies, the basic information database comprises basic information corresponding to each main body, and the basic information of the main bodies can be used as content for performing bottom-to-bottom reply on the main bodies. The basic information corresponding to the subject name in the basic information database can be acquired according to the subject name of the target subject node and fed back to the user terminal as the pocket bottom reply information.

The technical method can be applied to application scenes such as intelligent government affairs, intelligent city management, intelligent community, intelligent security protection, intelligent logistics, intelligent medical treatment, intelligent education, intelligent environmental protection and intelligent traffic, wherein the application scenes comprise the steps of acquiring reply information and carrying out intelligent reply, and therefore the construction of the intelligent city is promoted.

In the artificial intelligence-based intelligent association response method provided by the embodiment of the invention, the mining data information is subjected to structured processing according to structured processing rules to generate an association knowledge network and added to a knowledge graph, a target main body node matched with the received question information and historical response information associated with the question information is obtained from the knowledge graph, if the target main body node and the question information meet preset response conditions, data of the target main body node is obtained and fed back to a user terminal as first response information, and if the preset response conditions are not met, node question information can be generated or bottom-of-pocket response information can be obtained and fed back to the user terminal. By the method, the matched target subject node in the knowledge graph can be obtained based on the question information and the historical response information associated with the question information, intelligent association response is carried out based on the data of the target subject node, and the accuracy of response to the complex question information can be improved.

The embodiment of the invention also provides an intelligent association response device based on artificial intelligence, which is used for executing any embodiment of the intelligent association response method based on artificial intelligence. Specifically, referring to fig. 9, fig. 9 is a schematic block diagram of an intelligent association response apparatus based on artificial intelligence according to an embodiment of the present invention. The artificial intelligence based intelligent association responding apparatus may be configured in the management server 10.

As shown in fig. 9, the artificial intelligence based intelligent relevance responding apparatus 100 includes a mined data information structuring processing unit 110, a relevance subject node obtaining unit 120, a target subject node obtaining unit 130, a judging unit 140, a first response information feedback unit 150, a child node judging unit 160, a node question information transmitting unit 170, and a bibliography response information feedback unit 180.

And the mining data information structuring processing unit 110 is configured to, if the input mining data information is received, perform structuring processing on the mining data information according to preset structuring processing rules to generate an associated knowledge network, and add the associated knowledge network to a pre-stored knowledge graph.

In one embodiment, the artificial intelligence based intelligent relevance responding apparatus 100 further comprises sub-units: the device comprises a data mining program configuration unit and a mining data information acquisition unit.

The data mining program configuration unit is used for receiving mining configuration information input by an administrator and configuring a pre-stored data mining program according to the mining configuration information to obtain a configured data mining program; and the mining data information acquisition unit is used for executing the configured data mining program to mine the mining data information.

In one embodiment, the mining data information structuring processing unit 110 includes sub-units: the system comprises a main body matching unit, an associated knowledge network acquisition unit and an associated knowledge network storage unit.

The main body matching unit is used for acquiring a main body corresponding to each piece of data in the mining data information according to the main body keyword information; the association knowledge network acquisition unit is used for performing structural association on the data of each main body according to the main body structural relationship information to generate an association knowledge network containing all the main bodies; and the associated knowledge network storage unit is used for adding the associated knowledge network to the knowledge graph for storage.

An associated subject node obtaining unit 120, configured to receive the question information from the user terminal, and obtain, from the knowledge graph, an associated subject node that matches the question information and pre-stored historical question-answer information.

In an embodiment, the association subject node obtaining unit 120 includes sub-units: an associated questioning information acquisition unit and a node matching unit.

The relevant question information acquisition unit is used for acquiring multiple rounds of historical question information which is the same as the application scene information of the question information from the historical question answering information as relevant question information; and the node matching unit is used for acquiring the associated main body node of the main body node matched with the associated questioning information in the knowledge graph.

A target subject node obtaining unit 130, configured to obtain, according to a preset matching model, one target subject node that matches the question information from the associated subject node.

In one embodiment, the target subject node obtaining unit 130 includes sub-units: the system comprises a combined information acquisition unit, a question feature vector acquisition unit, a node data feature vector acquisition unit, a calculation unit, a similarity acquisition unit and a target subject node determination unit.

The combined information acquisition unit is used for combining the question information and the associated question information to obtain combined information; the question feature vector acquisition unit is used for converting the combined information according to a preset vocabulary processing rule to obtain a corresponding question feature vector; the node data characteristic vector acquisition unit is used for converting the node data of the associated main body node according to the vocabulary processing rule to obtain a corresponding node data characteristic vector; the computing unit is used for respectively computing the question feature vector and the node data feature vector according to the neural network to obtain corresponding question output information and node data output information; the similarity obtaining unit is used for calculating the similarity between the question output information and each node data output information according to the similarity calculation formula; and the target subject node determining unit is used for acquiring the associated subject node of the node data feature vector with the highest similarity as the target subject node.

A determining unit 140, configured to determine whether the target subject node and the question information satisfy a preset reply condition.

In one embodiment, the determining unit 140 includes sub-units: the device comprises a similarity judging unit, a matching judging unit and a judging result acquiring unit.

A similarity determination unit, configured to determine whether a similarity between the subject node data of the target subject node and the combined information is greater than a similarity threshold in the preset reply condition; a matching judgment unit, configured to judge whether the question information matches node constraint information corresponding to the main node data if the similarity between the main node data and the combined information is greater than the similarity threshold; and the judgment result acquisition unit is used for judging that the target main body node and the question information meet the preset reply condition if the question information is matched with the node limit information of the main body node.

A first reply information feedback unit 150, configured to, if the target subject node and the question information satisfy a preset reply condition, obtain subject node data of the target subject node as first reply information and feed the first reply information back to the user terminal.

A child node determining unit 160, configured to determine whether the target subject node includes a child node if the target subject node and the question information do not satisfy a preset reply condition.

A node question information sending unit 170, configured to generate node question information according to the child node of the target subject node and send the node question information to the user terminal if the target subject node includes the child node, so as to obtain reply information fed back by the user terminal according to the node question information.

In one embodiment, the artificial intelligence based intelligent relevance responding apparatus 100 further comprises sub-units: a second reply information acquisition unit.

And the second reply information acquisition unit is used for acquiring second reply information matched with the question information and the reply information in the child nodes of the target main body node and feeding the second reply information back to the user terminal.

And a bottom-of-pocket reply information feedback unit 180, configured to, if the target subject node does not include a child node, obtain, from a pre-stored basic information database, basic information corresponding to the target subject node as bottom-of-pocket reply information and feed the bottom-of-pocket reply information back to the user terminal.

The intelligent association response device based on artificial intelligence provided by the embodiment of the invention applies the intelligent association response method based on artificial intelligence, carries out structural processing on mined data information according to structural processing rules to generate an association knowledge network and adds the association knowledge network into a knowledge graph, acquires a target main body node matched with received question information and historical response information associated with the question information from the knowledge graph, acquires data of the target main body node as first response information and feeds the first response information back to a user terminal if the target main body node and the question information meet preset response conditions, and can generate node question information or acquire bottom response information and feed the bottom response information back to the user terminal if the preset response conditions are not met. By the method, the matched target subject node in the knowledge graph can be obtained based on the question information and the historical response information associated with the question information, intelligent association response is carried out based on the data of the target subject node, and the accuracy of response to the complex question information can be improved.

The above-mentioned intelligent association reply means based on artificial intelligence may be implemented in the form of a computer program which can be run on a computer device as shown in fig. 10.

Referring to fig. 10, fig. 10 is a schematic block diagram of a computer device according to an embodiment of the present invention. The computer device may be a management server 10 for executing an artificial intelligence based intelligent association reply method to obtain reply information for intelligent reply.

Referring to fig. 10, the computer device 500 includes a processor 502, memory, and a network interface 505 connected by a system bus 501, where the memory may include a non-volatile storage medium 503 and an internal memory 504.

The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. The computer programs 5032, when executed, cause the processor 502 to perform an artificial intelligence based intelligent association response method.

The processor 502 is used to provide computing and control capabilities that support the operation of the overall computer device 500.

The internal memory 504 provides an environment for the execution of the computer program 5032 in the non-volatile storage medium 503, and when the computer program 5032 is executed by the processor 502, the processor 502 can be enabled to execute the artificial intelligence based intelligent association response method.

The network interface 505 is used for network communication, such as providing transmission of data information. Those skilled in the art will appreciate that the configuration shown in fig. 10 is a block diagram of only a portion of the configuration associated with aspects of the present invention and is not intended to limit the computing device 500 to which aspects of the present invention may be applied, and that a particular computing device 500 may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

The processor 502 is configured to run the computer program 5032 stored in the memory to implement the corresponding functions in the artificial intelligence based intelligent association response method.

Those skilled in the art will appreciate that the embodiment of a computer device illustrated in fig. 10 does not constitute a limitation on the specific construction of the computer device, and that in other embodiments a computer device may include more or fewer components than those illustrated, or some components may be combined, or a different arrangement of components. For example, in some embodiments, the computer device may only include a memory and a processor, and in such embodiments, the structures and functions of the memory and the processor are consistent with those of the embodiment shown in fig. 10, and are not described herein again.

It should be understood that, in the embodiment of the present invention, the Processor 502 may be a Central Processing Unit (CPU), and the Processor 502 may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components, and the like. Wherein a general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

In another embodiment of the invention, a computer-readable storage medium is provided. The computer readable storage medium may be a non-volatile computer readable storage medium. The computer readable storage medium stores a computer program, wherein the computer program, when executed by a processor, implements the steps included in the artificial intelligence based intelligence associative reply method described above.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described apparatuses, devices and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again. Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the examples described in connection with the embodiments disclosed herein may be embodied in electronic hardware, computer software, or combinations of both, and that the components and steps of the examples have been described in a functional general in the foregoing description for the purpose of illustrating clearly the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

In the embodiments provided by the present invention, it should be understood that the disclosed apparatus, device and method can be implemented in other ways. For example, the above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only a logical division, and there may be other divisions when the actual implementation is performed, or units having the same function may be grouped into one unit, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may also be an electric, mechanical or other form of connection.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention.

In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention essentially contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product stored in a computer-readable storage medium, which includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned computer-readable storage media comprise: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a magnetic disk, or an optical disk.

While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. An intelligent association reply method based on artificial intelligence is applied to a management server, the management server and at least one user terminal realize the transmission of data information through network connection, and the method is characterized by comprising the following steps:

2. The method of claim 1, wherein before receiving the inputted mining data information, performing a structuring process on the mining data information according to a preset structuring process rule to generate a related knowledge network, and adding the related knowledge network to a pre-stored knowledge graph, the method further comprises:

receiving mining configuration information input by an administrator, and configuring a pre-stored data mining program according to the mining configuration information to obtain a configured data mining program, wherein the mining configuration information comprises mining keyword information, time information and a mining address set;

and executing the configured data mining program to mine to obtain the mined data information.

3. The artificial intelligence based intelligent relevance responding method according to claim 1, wherein the structured processing rules comprise subject keyword information and subject structural relationship information, and the structuring the mining data information according to preset structured processing rules to generate a relevant knowledge network and adding the relevant knowledge network to a pre-stored knowledge graph comprises:

acquiring a main body corresponding to each piece of data in the mining data information according to the main body keyword information;

performing structural association on the data of each main body according to the main body structural relationship information to generate an associated knowledge network containing all the main bodies;

and adding the associated knowledge network into the knowledge graph for storage.

4. The method according to claim 1, wherein the obtaining of the associated subject nodes matching the question information and the pre-stored historical question-answer information from the knowledge-graph comprises:

acquiring multiple rounds of historical question information which is the same as the application scene information of the question information from the historical question-answering information as associated question information;

and acquiring the associated main body nodes of the main body nodes matched with the associated questioning information in the knowledge graph.

5. The artificial intelligence based intelligent relevance answering method according to claim 4, wherein the matching model includes vocabulary processing rules, neural networks and similarity calculation formulas, and the obtaining of a target subject node matching the question information from the relevance subject nodes according to a preset matching model includes:

combining the question information and the associated question information to obtain combined information;

converting the combined information according to a preset vocabulary processing rule to obtain a corresponding question feature vector;

converting the node data of the associated main body node according to the vocabulary processing rule to obtain a corresponding node data characteristic vector;

calculating the question feature vector and the node data feature vector according to the neural network respectively to obtain corresponding question output information and node data output information;

calculating the similarity between the question output information and each node data output information according to the similarity calculation formula;

and acquiring the associated main body node of the node data feature vector with the highest similarity as the target main body node.

6. The artificial intelligence based intelligent relevance responding method according to claim 5, wherein the preset responding condition includes a similarity threshold, and the determining whether the target subject node and the question information satisfy the preset responding condition includes:

judging whether the similarity between the main body node data of the target main body node and the combined information is greater than a similarity threshold value in the preset reply condition or not;

if the similarity between the main node data and the combined information is larger than the similarity threshold, judging whether the questioning information is matched with node limiting information corresponding to the main node data;

and if the question information is matched with the node limiting information of the main body node, judging that the target main body node and the question information meet the preset reply condition.

7. The artificial intelligence based intelligent correlation reply method according to claim 1, wherein after the node question generation information according to the child node of the target subject node is sent to the user terminal to obtain the reply information fed back by the user terminal according to the node question information, the method further comprises:

and acquiring second reply information matched with the question information and the reply information in the child nodes of the target main body node and feeding back the second reply information to the user terminal.

8. An intelligent association reply device based on artificial intelligence, comprising:

9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the artificial intelligence based intelligent association reply method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which, when executed by a processor, causes the processor to carry out an artificial intelligence based intelligence correlation reply method according to any one of claims 1 to 7.