CN110019709B - Automatic question and answer method for robot and automatic question and answer system for robot - Google Patents
Automatic question and answer method for robot and automatic question and answer system for robot Download PDFInfo
- Publication number
- CN110019709B CN110019709B CN201711173973.5A CN201711173973A CN110019709B CN 110019709 B CN110019709 B CN 110019709B CN 201711173973 A CN201711173973 A CN 201711173973A CN 110019709 B CN110019709 B CN 110019709B
- Authority
- CN
- China
- Prior art keywords
- request information
- knowledge
- question
- reasoning
- instance
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000004364 calculation method Methods 0.000 claims abstract description 27
- 230000008569 process Effects 0.000 claims abstract description 23
- 238000004590 computer program Methods 0.000 claims description 2
- 230000014509 gene expression Effects 0.000 description 19
- 238000005516 engineering process Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 4
- 230000003993 interaction Effects 0.000 description 4
- 239000004065 semiconductor Substances 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 239000010881 fly ash Substances 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 230000037396 body weight Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a robot automatic question and answer method, a device, a server and a storage medium, wherein the method comprises the following steps: receiving initial request information of a user, wherein the initial request information comprises a first instance and N attribute information, and N is more than or equal to 2; searching knowledge points corresponding to the initial request information in a knowledge base through similarity calculation; when the corresponding knowledge points are not found in the knowledge base, carrying out reasoning processing to obtain the request information after reasoning, wherein the request information after reasoning comprises a second instance and M pieces of attribute information, and M is smaller than or equal to N-1; searching knowledge points corresponding to the inferred request information in a knowledge base through similarity calculation; when the corresponding knowledge points are not found in the knowledge base, the above reasoning and calculating processes are repeated until the corresponding knowledge points are found in the knowledge base. The invention realizes that the answers of complex questions are found on the basis of not changing the knowledge points stored in the knowledge base.
Description
Technical Field
The embodiment of the invention relates to a man-machine interaction technology, in particular to a robot automatic question-answering method, a robot automatic question-answering system, a server and a storage medium.
Background
Human-computer interaction is a science of studying the interaction relationship between a system and a user, and the system can be various machines or computerized systems and software. The robot automatic question-answering system is an artificial intelligent system developed by means of man-machine interaction technology, such as an intelligent customer service system, a voice control system and the like.
The knowledge base is the basis of the robotic automated question-answering system in which information is efficiently organized for retrieval and utilization. A plurality of knowledge points are stored in the knowledge base, each knowledge point including one or more questions and corresponding answers. When a user inputs request information, calculating the semantic similarity between the request information and the questions in the knowledge base, and if the questions with the similarity larger than a first preset threshold exist, returning answers corresponding to the questions with the maximum similarity to the user.
In general, the knowledge base stores simpler questions, such as "Liu Dehua wife is" if the user's request information is also a simpler question, and the corresponding answer can be directly obtained through similarity calculation, but if the user's request information is more complex, including a plurality of simple questions, for example: the first girl of the present girl of the former doctor of XXX is who the girl of the first husband is, and the corresponding answer cannot be obtained from the knowledge base.
Disclosure of Invention
In view of this, the embodiments of the present invention provide a method for automatic question-answering by a robot, a system for automatic question-answering by a robot, a server, and a storage medium, so as to find an answer to a complex question on the basis of not changing knowledge points stored in a knowledge base, thereby improving accuracy of question-answering.
The embodiment of the invention provides a robot automatic question-answering method, which comprises the following steps:
receiving initial request information of a user, wherein the initial request information comprises a first instance and N attribute information, and N is more than or equal to 2;
Searching knowledge points corresponding to the initial request information in a knowledge base through similarity calculation;
when the corresponding knowledge points are not found in the knowledge base, carrying out reasoning processing to obtain the request information after reasoning, wherein the request information after reasoning comprises a second instance and M pieces of attribute information, and M is smaller than or equal to N-1;
searching knowledge points corresponding to the inferred request information in a knowledge base through similarity calculation;
when the corresponding knowledge points are not found in the knowledge base, repeating the reasoning and calculating processes until the corresponding knowledge points are found in the knowledge base, wherein the knowledge points comprise answers;
and sending the answers of the searched knowledge points to the user.
Optionally, the performing inference processing to obtain the inferred request information includes:
carrying out reasoning processing according to a pre-established knowledge graph to obtain the request information after reasoning;
the knowledge graph comprises a plurality of examples in a knowledge base and association relations among the examples.
Optionally, the knowledge points in the knowledge base include questions and answers, and when the questions include a first instance and the answers include a second instance, the knowledge points further include association relations between the first instance and the second instance.
Optionally, the performing inference processing according to a pre-established knowledge graph to obtain the inferred request information includes:
Calculating the similarity of the initial request information and the problems in the knowledge base, wherein the initial request information comprises a first instance;
Selecting a set of questions with the similarity greater than a second preset threshold as an extended greeting selection set of the initial request information, wherein the extended greeting selection set comprises a plurality of candidate questions, and the candidate questions in the extended question candidate set comprise the first instance;
selecting a candidate question to be processed from the extended question candidate set;
When the candidate questions to be processed cannot consume the words in the initial request information in turn from left to right, re-selecting one candidate question to be processed from the extended question candidate set until the selected candidate question to be processed can consume the words in the initial request information in turn from left to right, and taking the selected candidate question to be processed as a target question;
And obtaining an answer corresponding to the target candidate question, and when the answer references a second instance, replacing the word consumed by the target candidate question in the initial request information with the second instance to obtain the inferred request information.
The embodiment of the invention also provides a system for automatically asking and answering the robot, which comprises the following steps:
the user request receiving module is used for receiving initial request information of a user, wherein the initial request information comprises a first instance and N attribute information, and N is more than or equal to 2;
The first knowledge point searching module is used for searching knowledge points corresponding to the initial request information in a knowledge base through similarity calculation;
the reasoning module is used for carrying out reasoning processing when the corresponding knowledge point is not found in the knowledge base so as to obtain the request information after reasoning, wherein the request information after reasoning comprises a second instance and M pieces of attribute information, and M is smaller than or equal to N-1;
the second knowledge point searching module is used for searching knowledge points corresponding to the inferred request information in a knowledge base through similarity calculation, and the knowledge points comprise answers;
The knowledge point determining module is used for continuously repeatedly triggering the reasoning module and the second knowledge point searching module to conduct the reasoning and calculating process when the corresponding knowledge point is not found in the knowledge base until the corresponding knowledge point is found in the knowledge base;
and the output module is used for sending the answers of the searched knowledge points to the user.
Optionally, the reasoning module is specifically configured to:
When the corresponding knowledge points are not found in the knowledge base, carrying out reasoning processing according to the pre-established knowledge graph to obtain the request information after reasoning;
the knowledge graph comprises a plurality of examples in a knowledge base and association relations among the examples.
Optionally, the knowledge points in the knowledge base include questions and answers, and when the questions include a first instance and the answers include a second instance, the knowledge points further include association relations between the first instance and the second instance.
Optionally, the reasoning module includes:
A similarity calculation unit, configured to calculate a similarity between the initial request information and a problem in a knowledge base, where the initial request information includes a first instance;
An extended greeting selection determining unit, configured to select, as an extended greeting selection of the initial request information, a set of questions having the similarity greater than a second preset threshold, where the extended greeting selection includes a plurality of questions candidates, and the questions in the extended question candidate set include the first instance;
a candidate question selecting unit for selecting one candidate question from the extended question candidate set;
A target candidate question determining unit, configured to, when the candidate questions to be processed cannot consume the words in the initial request information sequentially from left to right, reselect one candidate question to be processed from the extended question candidate set until the selected candidate question to be processed can consume the words in the initial request information sequentially from left to right, and take the selected candidate question to be processed as a target question;
And the post-reasoning request information determining unit is used for acquiring an answer corresponding to the target candidate question, and when the answer references a second instance, replacing the word consumed by the target candidate question in the initial request information with the second instance to obtain the post-reasoning request information.
The embodiment of the invention also provides a server, which comprises:
one or more processors;
Storage means for storing one or more programs,
When the one or more programs are executed by the one or more processors, the one or more processors implement the method for automatic question-answering of a robot.
The embodiment of the invention also provides a computer storage medium, on which a computer program is stored, which when being executed by a processor, realizes the automatic question-answering method of the robot.
According to the technical scheme, when the knowledge points corresponding to the initial request information cannot be found in the knowledge base, the reasoning processing is carried out to obtain the request information after reasoning, the knowledge points corresponding to the request information after reasoning are found in the knowledge base through similarity calculation, and when the corresponding knowledge points are not found in the knowledge base, the reasoning and calculating processes are repeated until the corresponding knowledge points are found in the knowledge base, so that answers to complex questions are found on the basis of not changing the knowledge points stored in the knowledge base, and on the premise of not changing the storage content of the knowledge base, the accuracy of questions and answers is improved, and the storage space is saved.
Drawings
Fig. 1 is a flowchart of a method for automatic question-answering by a robot according to an embodiment of the present invention;
Fig. 2 is a flowchart of an inference process according to a pre-established knowledge graph in the method for automatic question answering of a robot according to the first embodiment of the present invention;
fig. 3 is a schematic structural diagram of a robot automatic question-answering system according to a second embodiment of the present invention;
fig. 4 is a schematic structural diagram of a server according to a third embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to the drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting thereof. It should be further noted that, for convenience of description, only some, but not all of the matters related to the present invention are shown in the accompanying drawings.
In order to facilitate understanding of the scheme of the embodiment of the invention, the following simple description is firstly made on the basic knowledge of the automatic robot question-answering system:
1. Knowledge points
The most primitive and simplest form of the basic knowledge points in the knowledge base is the FAQ commonly used in normal times, and the general form is a "question-answer" pair. For example, "tariffs for color ring" is a clear-expressed standard question description. "question" here should not be interpreted narrowly as "query" but rather broadly as an "input" with a corresponding "output". For example, for semantic recognition of a control system, an instruction by the user, such as "turn on radio", should also be understood as a "question", in which case the corresponding "answer" may be a call to a control program for executing the respective control.
When a user inputs to the machine, the most ideal case is that the user uses a standard question, and the intelligent semantic recognition system of the machine can immediately understand the meaning of the user. However, users often do not use a question of standards, but rather are some variant of a question of standards. For example, if the standard format of station switching for radio is "change station", then the user may use the command "switch station" and the machine needs to be able to recognize that the user is expressing the same meaning.
Thus, for intelligent semantic recognition, there is a need in the knowledge base for an extension of the question, which is slightly different from the expression form of the question, but which expresses the same meaning.
Thus, the knowledge base includes a plurality of knowledge points, each knowledge point including questions and answers, the questions including a standard question and a plurality of extended questions.
The questions in the knowledge points are typically represented in the form of semantic expressions, the symbols in the semantic expressions being as follows:
1.1.1 representation of word class ([ ])
To distinguish words from parts of speech in an expression, it is specified that parts of speech must appear in brackets "[ ]", the parts of speech appearing in brackets are typically "narrow parts of speech", but "broad parts of speech" may also be supported by configuring system parameters. The following are examples of some simple expressions: [ how to open a flyash ], [ introduction ] [ multimedia message ], [ method of logging in a flyash ], [ how to charge ] a caller alert ].
1.1.2 Representation of the relationship (|)
The parts of speech in brackets may appear multiple times in "or" relationships, which are calculated separately in an "expanded" manner when calculating similarity. "expansion" is mainly a process of expanding a semantic expression into a plurality of simple expressions according to the meaning of "or". Such as: the [ method|step ] of [ color ring ] [ on ] can be expanded into two simple semantic expressions of [ step ] "of [ color ring ] [ on ] and [ method ]" of [ color ring ] [ on ]. Examples of such semantic expressions are as follows: how query know PUK code, unsubscribe cancel shut down stop IP 17951 domestic long distance offer package.
1.1.3 Optional representation (
The parts of speech in brackets may be added at the end "? "means that the relationship may or may not occur, i.e., that unnecessary relationships, the parts of speech of which are also calculated separately in a" unrolled "manner when calculating the similarity. "expansion" is essentially the process of expanding a semantic expression that contains an unnecessary part of speech (or "or combination" of parts of speech) into two simple semantic expressions that contain and do not contain this part of speech.
Such as: what are? The method can be developed into two simple semantic expressions of 'introduction' of 'military column' content 'and' introduction 'of' military column 'content'.
2. Body
Ontologies (ontologies) are formalized representations of a set of concepts and their relationships to each other within a particular domain, and are abstractions of a domain. We use an ontology to obtain knowledge of a domain, the ontology describing concepts of the domain, and the relationships between these concepts.
2.1 Construction of the body
The ontology comprises an ontology class and an instance, the ontology class defines the structure of the corresponding instance, and the instance is an individual object conforming to the definition of the ontology class.
2.1.1 Ontology classes (Ontology Class)
In the real world there are often objects belonging to the same class (e.g. a car is only one of many cars in the world) and there are also many different objects sharing the same features. The same features of these objects can be used to build a prototype for them. An Ontology Class is a definition of this prototype, which is an abstraction of all individuals that have or share common characteristics. The characteristics of an ontology Class are called Class properties (Class Property), for example, the "people" ontology Class has properties such as "name", "age" and "body weight". Class attributes consist of an attribute name and a set of semantic expressions. In an ontology class, we can describe the semantics expressed by attributes in a set of semantic expressions, the semantic expressions of class attributes consisting of instantiations, parts of speech, grammar symbols, the instantiations representing parts of speech or semantic blocks describing individuals.
2.1.2 Example (Instance)
An instance is the product of instantiation of an ontology class, representing a specific individual object that conforms to the definition of the ontology class. An instance consists of an instance name, instance semantics, attributes, and attribute values. Instance semantics: instance semantics are semantic expressions that describe instance semantics that replace "instantiations" in the ontology class attribute semantic expressions in instantiating an ontology class, thereby generating semantics for instance attributes.
Instance attributes: instance attributes include general attributes and custom attributes. The general attribute is an attribute defined in the body class of the instantiation instance, for example, the instance of the 'Santana car' is instantiated by the body class of the 'car', and then the attributes such as the 'wheelbase', 'weight', and 'displacement' defined in the body class of the 'car' are the general attributes. Custom attributes are feature descriptions that are custom outside of general attributes that are tailored to an individual for more comprehensive description of that individual.
Attribute value: the attribute value, i.e., a specific description of an attribute, is used to express exactly the quality or quantity of the attribute of an instance. For example, the "male" is the attribute value of the "sex" attribute in the example of "Zhang Sano".
In the following embodiments of the present invention, each knowledge point stores only one attribute information between two instances, i.e., a simple knowledge point.
Example 1
Fig. 1 is a flowchart of a robot automatic question-answering method provided in an embodiment of the present invention, where the embodiment is applicable to searching for a knowledge point corresponding to more complex user request information, the method may be performed by a robot automatic question-answering device, and the device may be implemented by software and/or hardware, and may be generally integrated on a server or other devices, and the method specifically includes the following steps:
step 110, receiving initial request information of a user, wherein the initial request information comprises a first instance and N pieces of attribute information, and N is greater than or equal to 2.
Initial request information of a user can be received through the interactive interface. The initial request information may be request information in a voice form or request information in a text form. When the initial request information is request information in a voice form, the request information needs to be converted into request information in a text form by voice recognition.
The embodiment mainly solves the problem of automatic question and answer of initial request information comprising a first instance and a plurality of attribute information. Such as: if the girl of the first husband of the former girl of XXX is who, XXX is the first instance in the initial request information, and the former, the present girl, the first husband and the girl are four attribute values.
And 120, searching knowledge points corresponding to the initial request information in a knowledge base through similarity calculation.
And calculating the semantic similarity between the initial request information and the problems stored in the knowledge base, and comparing the calculated similarity with a first preset threshold value to find a knowledge point corresponding to the initial request information. Wherein, the questions in the knowledge base comprise standard questions and corresponding extension questions, and the extension questions have different expression forms from the standard questions, but have the same expression meaning.
Step 130, judging whether a knowledge point corresponding to the initial request information is found in the knowledge base, if not, executing step 140, and if so, executing step 170.
If the problem that the similarity is larger than the first preset threshold exists, the problem with the largest similarity can be selected, and the knowledge point comprising the problem is determined to be the knowledge point corresponding to the initial request information; if the problem that the similarity is larger than the first preset threshold value does not exist, determining that a knowledge point corresponding to the initial request information cannot be found in the knowledge base.
And 140, performing reasoning processing to obtain the request information after reasoning, wherein the request information after reasoning comprises a second instance and M pieces of attribute information, and M is smaller than or equal to N-1.
The present embodiment reduces at least one attribute information, i.e., at least one attribute relationship between the first instance and the second instance, while changing the first instance to the second instance through one inference process. The number of attribute information that is specifically reduced is determined based on knowledge points that are specifically stored in the knowledge base.
When the corresponding knowledge points cannot be found in the knowledge base, the request information may be complex, and the request information may be processed by reasoning to obtain the request information after reasoning.
The reasoning processing is performed to obtain the request information after reasoning, which comprises the following steps:
carrying out reasoning processing according to a pre-established knowledge graph to obtain the request information after reasoning;
the knowledge graph comprises a plurality of examples in a knowledge base and association relations among the examples.
In a field or a specific item, association relations exist among all the examples in the field or the specific item, and a knowledge graph is formed. When the initial request information of the user is complex, the reasoning process can be performed according to the pre-established knowledge graph. Specifically, the knowledge graph may be in the form of a table, for example, the knowledge points corresponding to the example 1 include 1.1, 1.2, 1.3, etc., while the answer of the knowledge point 1.1 references another example 2, the knowledge points corresponding to the example 2 include 2.1, 2.2, etc., and the answer of the knowledge point 2.2 references the example 1. By constructing the association relationship between the instances, reasoning can be rapidly realized.
Generally, when the initial request information includes the first instance, if the initial request information is to obtain an attribute value of an attribute of the first instance, such as who is the former of XXX, that is, XXX is the first instance, the former is an attribute of the first instance, and the specific answer is an attribute value of the attribute, then the initial request information may be considered as a simple problem; and if the initial request information is a second attribute value of a second attribute of the first instance, that is, an attribute in which at least two attributes are included and the next attribute is an attribute corresponding to an attribute value of the previous attribute, the initial request information may be regarded as a complicated question, such as who the girl of the first husband of the former girl of XXX is, including the first instance "XXX", the first attribute "former of the first instance", the second attribute "present girl" of the corresponding first attribute value, the third attribute "first husband" of the corresponding second attribute value, and the fourth attribute "girl" of the corresponding third attribute value, the final answer should be the fourth attribute value corresponding to the fourth attribute, in the reasoning process, when the first reasoning is carried out, a first attribute value is obtained, then initial request information is converted into request information after the reasoning, namely, the first girl of the current girl of the first attribute value is who, a second reasoning is needed to be carried out, a second attribute value is obtained, then the initial request information is converted into request information after the two reasoning, namely, the first girl of the second attribute value is who, a third reasoning is needed to be carried out, a third attribute value is obtained, then the initial request information is converted into request information after the three reasoning, namely, the third girl of the third attribute value is who, and the problem becomes a simple problem, and corresponding answers can be directly obtained in a knowledge base.
Optionally, the knowledge points in the knowledge base include a question and an answer, and when the question includes a first instance and the answer includes a second instance, the knowledge points further include an association relationship between the first instance and the second instance, where the association relationship indicates that the answer of the question including the first instance references the second instance. Table 1 is an example of a storage structure of a knowledge point in a knowledge base, and as shown in table 1, the question "who is the producer of P9" corresponds to the answer "how is technology limited", and since "how is P9" is an example and "how is technology limited" is an example, an association relationship between the first example and the second example is also given in the knowledge point, that is, the answer corresponding to the question including the first example "how is P9" refers to the second example "how is technology limited".
For example, table 1 is two knowledge points in a knowledge base, the user initial request information is "who is the president of the producer of P9", the knowledge base does not directly store the answer corresponding to the initial request information, and the reasoning process flow is entered: obtaining a plurality of candidate questions of the initial request information according to similarity calculation, selecting one candidate question for processing, wherein the selected candidate question is: [ Hua is P9] [ manufacturer ] [ which of the bits is? Judging whether the candidate question can consume the words in the initial request information sequentially from left to right, acquiring an answer corresponding to the candidate question if the judging result is that the candidate question can consume sequentially, referring to another example of 'Hua as technology Limited company' from the table 1, replacing the consumed words in the initial request information with the answer, and obtaining the inferred request information 'Hua as who is a president of the technology Limited company'.
Table 1 storage structure examples of knowledge points in a knowledge base
Fig. 2 is a flowchart of an inference process according to a pre-established knowledge graph in the automatic robot question-answering method provided by the embodiment of the present invention, and as shown in fig. 2, the inference process is performed according to the pre-established knowledge graph to obtain the request information after the inference, including:
Step 141, calculating the similarity between the initial request information and the questions in the knowledge base, wherein the initial request information comprises a first instance.
Since the problems in the initial request information and the knowledge base are both represented by text, semantic similarity is calculated when calculating the similarity. Semantic similarity calculation refers to that given two texts (generally character strings), the similarity magnitude of the two texts is measured through an algorithm, and the general calculation result is a value between 0 and 1. Common methods for semantic similarity computation are neural network-based computation, typically DSSM (Deep Structured Semantic Models, deep structure semantic model) and CLSM (conditional LATENT SEMANTIC model), convolution latent semantic model. The DSSM principle is that a deep neural network is used for expressing problems and titles into low latitude semantic vectors through massive click exposure logs of the problems and titles in a search engine, the distance between the two semantic vectors is calculated through cosine distances, and finally a semantic similarity model is trained. CLSM is also known as CNN-DSSM, i.e. the deep neural network in DSSM is replaced with a convolutional neural network.
The first instance may be the first word in the initial request message, and of course, the first instance may be elsewhere in the initial request message.
Step 142, selecting the set of questions with the similarity greater than a second preset threshold as an extended greeting selection set of the initial request information, where the extended greeting selection set includes a plurality of candidate questions, and the candidate questions in the extended question candidate set include the first instance.
Wherein the second preset threshold is smaller than the first preset threshold.
The candidate questions in the extended question candidate set include a first instance to facilitate an inference process based on a pre-established knowledge graph. The candidate questions are simple questions similar to the initial request information in the knowledge base, so that answers corresponding to the candidate questions replace words consumed by the candidate questions in turn from left to right in the initial request information in the reasoning process, and the deduced request information is obtained, so that the initial request information is simplified.
And step 143, selecting a candidate question to be processed from the extended question candidate set.
In this embodiment, a candidate with the greatest similarity may be selected from the extended question candidate set, and used as a candidate to be processed. It should be noted that, the present invention may also arbitrarily select one candidate from the extended question candidate set as a candidate to be processed.
Step 144, determining whether the candidate question can consume the words in the initial request information sequentially from left to right, if not, returning to execute step 143 to reselect one candidate question from the extended question candidate set, and if so, executing step 145.
Wherein the word in the consumption candidate question is the same as or similar to the semantic meaning of the word in the initial request information.
And if the candidate question to be processed cannot consume the words in the initial request information from left to right in sequence, deleting the candidate question to be processed from the extended greeting selection set, and executing step 143, and re-selecting one candidate question to be processed from the extended question candidate set until the selected candidate question can consume the words in the initial request information from left to right in sequence. Wherein, the candidate question to be processed sequentially consumes the words in the initial request information from left to right refers to sequentially consuming part of the words (including the first instance and the attribute thereof) in the initial request information from left to right, and does not need to consume all the words. It should be noted that the consumed words may not include the auxiliary words.
And 145, taking the selected candidate question to be processed as a target candidate question.
And step 146, obtaining an answer corresponding to the target candidate question, and when the answer refers to a second instance, replacing the word consumed by the target candidate question in the initial request information with the second instance to obtain the inferred request information.
Preferably, selecting a candidate question to be processed from the extended question candidate set includes:
and selecting a candidate question with the highest similarity with the initial request information from the expansion question candidate set as a candidate question to be processed.
When the candidate question to be processed is selected from the extended question candidate set, the candidate question with the highest similarity is selected so as to quickly realize reasoning and quickly obtain knowledge points corresponding to the initial request information of the user.
And step 150, searching knowledge points corresponding to the inferred request information in a knowledge base through similarity calculation.
And calculating the semantic similarity of the inferred request information and the problems stored in the knowledge base, and comparing the calculated similarity with a first preset threshold value to find knowledge points corresponding to the inferred request information.
Step 160, judging whether the knowledge point corresponding to the request information after reasoning is found in the knowledge base, if not, returning to execute step 140, and if so, executing step 170.
If the problem that the similarity is larger than the first preset threshold exists, the problem with the maximum similarity can be selected, and the knowledge point comprising the problem is determined to be the knowledge point corresponding to the request information after reasoning; if the problem that the similarity is larger than the first preset threshold value does not exist, determining that knowledge points corresponding to the inferred request information cannot be found in the knowledge base. And when the knowledge points corresponding to the inferred request information cannot be found in the knowledge base, returning to the step 140, and continuing to conduct inference processing on the inferred request information in the manner described above until the corresponding knowledge points are found in the knowledge base.
And step 170, returning the answers in the searched knowledge points to the user.
The return mode can be text or voice, if the answer is voice, the answer in the searched text form is converted into the voice form, and the answer in the voice form is returned to the user.
The case of obtaining an answer based on one inference, for example, the initial request information of the user is "who is the president of the P9 producer? Calculating the semantic similarity of the initial request information and the problems in the knowledge base, wherein the calculated similarity is smaller than a first preset threshold value, and if the knowledge points corresponding to the initial request information cannot be found in the knowledge base, the reasoning process based on the knowledge map is carried out: firstly, calculating semantic similarity between initial request information and questions in a knowledge base, wherein the initial request information comprises a first example 'Hua is P9', selecting a set of questions with similarity larger than a second preset threshold value as an extended greeting selection set of the initial request information, selecting a candidate with highest similarity with the initial request information from the extended question candidate set as a candidate to be processed, and selecting the candidate to be processed as the candidate to be processed: [ Hua is P9] [ manufacturer ] [ which of the bits is? The candidate question to be processed can consume words in the initial request information from left to right in turn: the question is P9 and the producer, the candidate question to be processed is taken as a target candidate question, the question in the knowledge point corresponding to the target candidate question is "who the producer of the question is P9", the answer refers to a second example of "Hua is a technology limited company", the answer corresponding to the target candidate question is substituted for the word consumed by the candidate question in the initial request information, and the inferred request information "who is the president of the technology limited company" is obtained; through calculating the semantic similarity between the inferred request information and the questions in the knowledge base, the problem of finding that the similarity exceeds a first preset threshold is 'who is the president of the technology limited company', and the corresponding answer is 'any non-Mr.'; the answer is returned to the user.
Based on the situation that the answer is obtained through two times of reasoning, for example, the initial request information of the user is "what time is established by the manufacturer of the CPU model of P9", the semantic similarity between the initial request information and the questions in the knowledge base is calculated, the calculated similarity is smaller than a first preset threshold value, the fact that knowledge points corresponding to the initial request information cannot be found in the knowledge base is indicated, and a first reasoning flow based on the knowledge map is entered: firstly, calculating semantic similarity between initial request information and questions in a knowledge base, wherein the initial request information comprises a first example 'Hua is P9', selecting a set of questions with similarity larger than a second preset threshold value as an extended greeting selection set of the initial request information, selecting a candidate with highest similarity with the initial request information from the extended question candidate set as a candidate to be processed, and selecting the candidate to be processed as the candidate to be processed: what [ Hua is P9] [ CPU model ]? The candidate question to be processed can consume words in the initial request information from left to right in turn: the question is P9 (first example) and CPU model number (first attribute), the candidate question to be processed is taken as a target candidate question, the question in the knowledge point corresponding to the target candidate question is "what is the CPU model number of P9," the answer refers to a second example "Hai Si kylin 955," the answer corresponding to the target candidate question is substituted for the word consumed by the candidate question in the initial request information, and the inferred request information "what is the time the producer of Hai Si kylin 955" is; by calculating the semantic similarity between the inferred request information and the questions in the knowledge base, the similarity is smaller than a first preset threshold value, and the knowledge base still cannot find the corresponding knowledge points, at this time, the inferred request information "what time the manufacturer of the Hai Si kylin 955 holds" needs to be inferred, namely, the second inference flow based on the knowledge map is entered: firstly calculating semantic similarity between the inferred request information 'what the manufacturer of the Hai Si kylin 955 holds' and problems in a knowledge base, wherein the initial request information comprises a second example 'Hai Si kylin 955', selecting a set of problems with similarity larger than a second preset threshold value as an expanded greeting selection set of initial request information, and selecting a candidate with highest similarity with the initial request information from the expanded question candidate set as a candidate to be processed, wherein the selected candidate to be processed is: [ Hai Si kylin 955] [ manufacturer ] [ which of the? The candidate question to be processed can consume words in the initial request information from left to right in turn: a Hai Si kylin 955 (second example) and a producer (second attribute), taking the candidate question to be processed as a target candidate question, wherein the question in a knowledge point corresponding to the target candidate question is "who the producer of the Hai Si kylin 955 is", the answer refers to a third example "Hai Si semiconductor Limited company", the answer corresponding to the target candidate question is replaced with the word consumed by the candidate question in the inferred request information "what the producer of the Hai Si kylin 955 is, and the request information" what the time the Hai Si semiconductor Limited company is after the second inference is obtained "; by calculating the semantic similarity between the request information after the second reasoning and the problems in the knowledge base, finding out the problem that the similarity exceeds a first preset threshold value as 'what time is established by the Hai Si semiconductor limited company', and the corresponding answer is '10 months in 2004' of establishment time of the Hai Si semiconductor limited company; the answer is returned to the user.
The case of obtaining the answer based on three or more inferences can refer to the process of the above two inferences, and will not be described here again.
According to the technical scheme, when the knowledge points corresponding to the initial request information cannot be found in the knowledge base, reasoning processing is conducted to obtain the request information after reasoning, the knowledge points corresponding to the request information after reasoning are found in the knowledge base through similarity calculation, when the corresponding knowledge points are not found in the knowledge base, the reasoning and calculating processes are repeated until the corresponding knowledge points are found in the knowledge base, so that answers to complex questions are found on the basis of not changing the knowledge points stored in the knowledge base, and on the premise that the storage content of the knowledge base is not changed, the accuracy of questions and answers is improved, and storage space is saved.
Example two
Fig. 3 is a schematic structural diagram of a robot automatic question-answering system provided in the second embodiment of the present invention, where the system is suitable for searching for a knowledge point corresponding to relatively complex user request information, and the system may be implemented by software and/or hardware, and may be generally integrated on a server or other devices, and the system includes: the user request receiving module 310, the first knowledge point searching module 320, the reasoning module 330, the second knowledge point searching module 340, the knowledge point determining module 350 and the output module (not shown in the figure).
The user request receiving module 310 is configured to receive initial request information of a user, where the initial request information includes a first instance and N attribute information, and N is greater than or equal to 2;
A first knowledge point searching module 320, configured to search a knowledge point corresponding to the initial request information in a knowledge base through similarity calculation;
The reasoning module 330 is configured to perform reasoning processing when no corresponding knowledge point is found in the knowledge base, so as to obtain a request message after reasoning, where the request message after reasoning includes a second instance and M attribute information, and M is less than or equal to N-1;
a second knowledge point searching module 340, configured to search a knowledge point corresponding to the inferred request information in a knowledge base through similarity calculation;
The knowledge point determining module 350 is configured to, when no corresponding knowledge point is found in the knowledge base, repeatedly trigger the reasoning module and the second knowledge point finding module to perform the reasoning and the calculating processes until the corresponding knowledge point is found in the knowledge base;
and the output module 360 is used for sending the answer of the searched knowledge point to the user.
Optionally, the reasoning module is specifically configured to:
When the corresponding knowledge points are not found in the knowledge base, carrying out reasoning processing according to the pre-established knowledge graph to obtain the request information after reasoning;
the knowledge graph comprises a plurality of examples in a knowledge base and association relations among the examples.
Optionally, the knowledge points in the knowledge base include questions and answers, and when the questions include a first instance and the answers include a second instance, the knowledge points further include association relations between the first instance and the second instance.
Optionally, the reasoning module includes:
A similarity calculation unit, configured to calculate a similarity between the initial request information and a problem in a knowledge base, where the initial request information includes a first instance;
An extended greeting selection determining unit, configured to select, as an extended greeting selection of the initial request information, a set of questions having the similarity greater than a second preset threshold, where the extended greeting selection includes a plurality of questions candidates, and the questions in the extended question candidate set include the first instance;
a candidate question selecting unit for selecting one candidate question from the extended question candidate set;
A target candidate question determining unit, configured to, when the candidate questions to be processed cannot consume the words in the initial request information sequentially from left to right, reselect one candidate question to be processed from the extended question candidate set until the selected candidate question can consume the words in the initial request information sequentially from left to right, and take the selected candidate question to be processed as a target candidate question;
And the post-reasoning request information determining unit is used for acquiring an answer corresponding to the target candidate question, and when the answer references a second instance, replacing the word consumed by the target candidate question in the initial request information with the second instance to obtain the post-reasoning request information.
Optionally, the candidate query selection unit is specifically configured to:
and selecting a candidate question with the highest similarity with the initial request information from the expansion question candidate set as a candidate question to be processed.
The output module in this embodiment may be a voice output module, a display output module, or a combination of the two, which does not limit the protection scope of the present invention.
The specific working process of each module in this embodiment may refer to the first embodiment, and will not be described herein.
According to the technical scheme of the embodiment, initial request information of a user is received through a user request receiving module; the first knowledge point searching module searches knowledge points corresponding to the initial request information in a knowledge base through similarity calculation; when the corresponding knowledge points are not found in the knowledge base, the reasoning module conducts reasoning processing to obtain the request information after reasoning; the second knowledge point searching module searches knowledge points corresponding to the inferred request information in a knowledge base through similarity calculation; when the corresponding knowledge points are not found in the knowledge base, the knowledge point determining module continuously repeatedly triggers the reasoning module and the second knowledge point searching module to conduct reasoning and calculating until the corresponding knowledge points are found in the knowledge base, so that answers to complex questions are found on the basis of the knowledge points stored in the knowledge base without changing the knowledge base, the accuracy of questions and answers is improved on the premise of not changing the storage content of the knowledge base, and the storage space is saved.
Example III
Fig. 4 is a schematic structural diagram of a server according to a third embodiment of the present invention, as shown in fig. 4, where the server includes a processor 510, a storage device 520, an input device 530, and an output device 540; the number of processors 510 in the server may be one or more, one processor 510 being taken as an example in fig. 4; the processor 510, storage 520, input 530, and output 540 in the server may be connected by a bus or other means, for example in fig. 4.
The storage device 520 is used as a computer readable storage medium, and may be used to store a software program, a computer executable program, and modules, such as program instructions/modules corresponding to the method for establishing a knowledge base of automatic questioning and answering of a robot in the embodiment of the present invention (for example, the user request receiving module 310, the first knowledge point searching module 320, the reasoning module 330, the second knowledge point searching module 340, and the knowledge point determining module 350 in the apparatus for automatic questioning and answering of a robot). The processor 510 executes various functional applications and data processing of the terminal device by running software programs, instructions and modules stored in the storage device 520, that is, the method for implementing the automatic question-answering of the robot, which will not be described herein.
The storage 520 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, at least one application program required for functions; the storage data area may store data created according to the use of the terminal, etc. In addition, storage 520 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid-state storage device. In some examples, storage 520 may further include memory located remotely from processor 510, which may be connected to the terminal device via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input means 530 may be used to receive input numeric or character information and to generate key signal inputs related to user settings and function control of the terminal device. The output 540 may include a display device such as a display screen.
Example IV
A fourth embodiment of the present invention provides a storage medium containing computer-executable instructions for performing a method of robotic automatic question-answering when executed by a computer processor, the method comprising:
receiving initial request information of a user, wherein the initial request information comprises a first instance and N attribute information, and N is more than or equal to 2;
Searching knowledge points corresponding to the initial request information in a knowledge base through similarity calculation;
when the corresponding knowledge points are not found in the knowledge base, carrying out reasoning processing to obtain the request information after reasoning, wherein the request information after reasoning comprises a second instance and M pieces of attribute information, and M is smaller than or equal to N-1;
searching knowledge points corresponding to the inferred request information in a knowledge base through similarity calculation;
when the corresponding knowledge points are not found in the knowledge base, repeating the reasoning and calculating processes until the corresponding knowledge points are found in the knowledge base, wherein the knowledge points comprise answers;
and sending the answers of the searched knowledge points to the user.
Of course, the storage medium containing the computer executable instructions provided in the embodiment of the present invention is not limited to the above-mentioned method operations, but may also perform the related operations in the robot automatic question-answering method provided in the embodiment of the present invention, which are not described herein.
From the above description of embodiments, it will be clear to a person skilled in the art that the present invention may be implemented by means of software and necessary general purpose hardware, but of course also by means of hardware, although in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as a floppy disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a FLASH Memory (FLASH), a hard disk, or an optical disk of a computer, etc., and include several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments of the present invention.
It should be noted that, in the embodiment of the apparatus, each unit and module included are only divided according to the functional logic, but not limited to the above-mentioned division, so long as the corresponding function can be implemented; in addition, the specific names of the functional units are also only for distinguishing from each other, and are not used to limit the protection scope of the present invention.
Note that the above is only a preferred embodiment of the present invention and the technical principle applied. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, while the invention has been described in connection with the above embodiments, the invention is not limited to the embodiments, but may be embodied in many other equivalent forms without departing from the spirit or scope of the invention, which is set forth in the following claims.
Claims (4)
1. A method for automatic question answering by a robot, comprising:
receiving initial request information of a user, wherein the initial request information comprises a first instance and N attribute information, and N is more than or equal to 2;
Searching knowledge points corresponding to the initial request information in a knowledge base through similarity calculation;
when the corresponding knowledge points are not found in the knowledge base, carrying out reasoning processing to obtain the request information after reasoning, wherein the request information after reasoning comprises a second instance and M pieces of attribute information, and M is smaller than or equal to N-1;
searching knowledge points corresponding to the inferred request information in a knowledge base through similarity calculation;
when the corresponding knowledge points are not found in the knowledge base, repeating the reasoning and calculating processes until the corresponding knowledge points are found in the knowledge base, wherein the knowledge points comprise answers;
Sending the answers of the searched knowledge points to a user;
the reasoning processing is performed to obtain the request information after reasoning, which comprises the following steps:
carrying out reasoning processing according to a pre-established knowledge graph to obtain the request information after reasoning;
The knowledge graph comprises a plurality of examples in a knowledge base and association relations among the examples; the knowledge points in the knowledge base comprise questions and answers, and when the questions comprise a first instance and the answers comprise a second instance, the knowledge points further comprise association relations between the first instance and the second instance;
carrying out reasoning processing according to a pre-established knowledge graph to obtain the request information after reasoning, wherein the method comprises the following steps:
Calculating the similarity of the initial request information and the problems in the knowledge base, wherein the initial request information comprises a first instance;
Selecting a set of questions with the similarity greater than a second preset threshold as an extended greeting selection set of the initial request information, wherein the extended greeting selection set comprises a plurality of candidate questions, and the candidate questions in the extended question candidate set comprise the first instance;
selecting a candidate question to be processed from the extended question candidate set;
When the candidate questions to be processed cannot consume the words in the initial request information in turn from left to right, re-selecting one candidate question to be processed from the extended question candidate set until the selected candidate question to be processed can consume the words in the initial request information in turn from left to right, and taking the selected candidate question to be processed as a target question;
And obtaining an answer corresponding to the target candidate question, and when the answer references a second instance, replacing the word consumed by the target candidate question in the initial request information with the second instance to obtain the inferred request information.
2. A system for automated robotic question-answering, comprising:
the user request receiving module is used for receiving initial request information of a user, wherein the initial request information comprises a first instance and N attribute information, and N is more than or equal to 2;
The first knowledge point searching module is used for searching knowledge points corresponding to the initial request information in a knowledge base through similarity calculation;
the reasoning module is used for carrying out reasoning processing when the corresponding knowledge point is not found in the knowledge base so as to obtain the request information after reasoning, wherein the request information after reasoning comprises a second instance and M pieces of attribute information, and M is smaller than or equal to N-1;
the second knowledge point searching module is used for searching knowledge points corresponding to the inferred request information in a knowledge base through similarity calculation, and the knowledge points comprise answers;
The knowledge point determining module is used for continuously repeatedly triggering the reasoning module and the second knowledge point searching module to conduct the reasoning and calculating process when the corresponding knowledge point is not found in the knowledge base until the corresponding knowledge point is found in the knowledge base;
the output module is used for sending the answers of the searched knowledge points to the user;
the reasoning module is specifically configured to:
When the corresponding knowledge points are not found in the knowledge base, carrying out reasoning processing according to the pre-established knowledge graph to obtain the request information after reasoning;
The knowledge graph comprises a plurality of examples in a knowledge base and association relations among the examples; the knowledge points in the knowledge base comprise questions and answers, and when the questions comprise a first instance and the answers comprise a second instance, the knowledge points further comprise association relations between the first instance and the second instance;
The reasoning module comprises:
A similarity calculation unit, configured to calculate a similarity between the initial request information and a problem in a knowledge base, where the initial request information includes a first instance;
An extended greeting selection determining unit, configured to select, as an extended greeting selection of the initial request information, a set of questions having the similarity greater than a second preset threshold, where the extended greeting selection includes a plurality of questions candidates, and the questions in the extended question candidate set include the first instance;
a candidate question selecting unit for selecting one candidate question from the extended question candidate set;
A target candidate question determining unit, configured to, when the candidate questions to be processed cannot consume the words in the initial request information sequentially from left to right, reselect one candidate question to be processed from the extended question candidate set until the selected candidate question to be processed can consume the words in the initial request information sequentially from left to right, and take the selected candidate question to be processed as a target question;
And the post-reasoning request information determining unit is used for acquiring an answer corresponding to the target candidate question, and when the answer references a second instance, replacing the word consumed by the target candidate question in the initial request information with the second instance to obtain the post-reasoning request information.
3. A server, comprising:
one or more processors;
Storage means for storing one or more programs,
The one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of robotic automatic question and answer of claim 1.
4. A computer storage medium having stored thereon a computer program which when executed by a processor implements the method of automated questioning and answering of a robot as claimed in claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711173973.5A CN110019709B (en) | 2017-11-22 | 2017-11-22 | Automatic question and answer method for robot and automatic question and answer system for robot |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711173973.5A CN110019709B (en) | 2017-11-22 | 2017-11-22 | Automatic question and answer method for robot and automatic question and answer system for robot |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110019709A CN110019709A (en) | 2019-07-16 |
| CN110019709B true CN110019709B (en) | 2024-05-17 |
Family
ID=67185896
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711173973.5A Active CN110019709B (en) | 2017-11-22 | 2017-11-22 | Automatic question and answer method for robot and automatic question and answer system for robot |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110019709B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114490969B (en) * | 2021-12-29 | 2023-03-07 | 北京百度网讯科技有限公司 | Question and answer method and device based on table and electronic equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105608218A (en) * | 2015-12-31 | 2016-05-25 | 上海智臻智能网络科技股份有限公司 | Intelligent question answering knowledge base establishment method, establishment device and establishment system |
| WO2017076263A1 (en) * | 2015-11-03 | 2017-05-11 | 中兴通讯股份有限公司 | Method and device for integrating knowledge bases, knowledge base management system and storage medium |
| CN106776797A (en) * | 2016-11-22 | 2017-05-31 | 中国人名解放军理工大学 | A kind of knowledge Q-A system and its method of work based on ontology inference |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10394956B2 (en) * | 2015-12-31 | 2019-08-27 | Shanghai Xiaoi Robot Technology Co., Ltd. | Methods, devices, and systems for constructing intelligent knowledge base |
-
2017
- 2017-11-22 CN CN201711173973.5A patent/CN110019709B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017076263A1 (en) * | 2015-11-03 | 2017-05-11 | 中兴通讯股份有限公司 | Method and device for integrating knowledge bases, knowledge base management system and storage medium |
| CN105608218A (en) * | 2015-12-31 | 2016-05-25 | 上海智臻智能网络科技股份有限公司 | Intelligent question answering knowledge base establishment method, establishment device and establishment system |
| CN106776797A (en) * | 2016-11-22 | 2017-05-31 | 中国人名解放军理工大学 | A kind of knowledge Q-A system and its method of work based on ontology inference |
Non-Patent Citations (2)
| Title |
|---|
| 基于表示学习的知识库问答研究进展与展望;刘康;张元哲;纪国良;来斯惟;赵军;;自动化学报(06);全文 * |
| 网络教学平台中问答***的关键技术研究;曾庆鹏;吴水秀;;计算机与现代化(07);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110019709A (en) | 2019-07-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107885842B (en) | Intelligent question and answer method, device, server and storage medium | |
| US10490186B2 (en) | Parameter collection and automatic dialog generation in dialog systems | |
| CN109308357B (en) | Method, device and equipment for obtaining answer information | |
| CN106164869B (en) | Hybrid client/server architecture for parallel processing | |
| CN106649825B (en) | Voice interaction system and creation method and device thereof | |
| CN111708869B (en) | Processing method and device for man-machine conversation | |
| JP4890585B2 (en) | Dialog control system and program, and multidimensional ontology processing system and program | |
| CN111914179B (en) | Semantic-based fuzzy search method and device, storage medium and electronic equipment | |
| JP7488871B2 (en) | Dialogue recommendation method, device, electronic device, storage medium, and computer program | |
| CN110069612B (en) | Reply generation method and device | |
| CN110019729B (en) | Intelligent question-answering method, storage medium and terminal | |
| CN112035647B (en) | Question and answer method, device, equipment and medium based on man-machine interaction | |
| CN111191450A (en) | Corpus cleaning method, corpus entry device and computer-readable storage medium | |
| EP4060517A1 (en) | System and method for designing artificial intelligence (ai) based hierarchical multi-conversation system | |
| CN111400458A (en) | Automatic generalization method and device | |
| CN109829037A (en) | Method, system, server and the storage medium of intelligent automatic question answering | |
| CN111460120A (en) | Conversation management method, device, equipment and storage medium | |
| CN111026840A (en) | Text processing method, device, server and storage medium | |
| CN110377745B (en) | Information processing method, information retrieval device and server | |
| CN110019709B (en) | Automatic question and answer method for robot and automatic question and answer system for robot | |
| CN109656952B (en) | Query processing method and device and electronic equipment | |
| CN111506717A (en) | Question answering method, device, equipment and storage medium | |
| CN111737425B (en) | Response method, device, server and storage medium | |
| CN114579605A (en) | Table question-answer data processing method, electronic equipment and computer storage medium | |
| CN114756655A (en) | Data query method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |