CN113010632A - Intelligent question answering method and device, computer equipment and computer readable medium - Google Patents
Intelligent question answering method and device, computer equipment and computer readable medium Download PDFInfo
- Publication number
- CN113010632A CN113010632A CN201911328449.XA CN201911328449A CN113010632A CN 113010632 A CN113010632 A CN 113010632A CN 201911328449 A CN201911328449 A CN 201911328449A CN 113010632 A CN113010632 A CN 113010632A
- Authority
- CN
- China
- Prior art keywords
- query
- semantic
- question
- request
- knowledge graph
- 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.)
- Pending
Links
Images
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
-
- 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
-
- 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
-
- 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
-
- 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)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (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 present disclosure provides an intelligent question answering method, which includes: acquiring an external request, if the external request is an inquiry request, determining a conversation state, and performing semantic analysis on the inquiry request by using a preset semantic template to obtain semantic information; if the inquiry request relates to a preset field and the conversation state is a non-interruption state, respectively inquiring a knowledge graph and a question-answering engine according to semantic information, and generating an answer according to the inquiry result of the knowledge graph and the inquiry result of the question-answering engine; the embodiment of the disclosure can be applied to intelligent customer service, knowledge management and intelligent marketing, reduces the burden of agent customer service personnel and background operation and maintenance management personnel, improves the service quality and knowledge management level, and improves the marketing efficiency. The present disclosure also provides an intelligent question answering device, a computer device and a computer readable medium.
Description
Technical Field
The disclosure relates to the technical field of knowledge graphs, in particular to an intelligent question answering method, an intelligent question answering device, computer equipment and a computer readable medium.
Background
In recent years, with the progress of technology development, large telecom operators have started to build intelligent question-answering systems using natural language processing technology for handling the rapidly increasing customer service demands.
The intelligent question-answering system is an important direction for the application of natural language processing technology. From a technical point of view, the intelligent question-answering system can be divided into a search-type question-answering system and a generation-type dialogue system. The retrieval type question-answering system is based on a large-scale question-answering corpus and provides answers most relevant to the question sentences for the user through an information retrieval technology. The generative dialogue system is also based on large-scale corpora, but does not directly retrieve contents from a corpus, and learns to obtain a generative model from question to answer through statistical learning or even deep learning methods. The query-type question-answering system has good predictability, interpretability and maintainability, is convenient for large-scale commercial use, and can realize good performance under the condition that the question-answering corpus is abundant enough. The generative question-answering system can process more complex semantics, and has better intelligence and generalization, but the internal working mechanism of the model is difficult to explain, and a great amount of time, manpower and linguistic data resources are consumed for training and maintaining the model. Considering the compromise between performance and cost, most intelligent question-answering systems are retrieval type question-answering systems.
Disclosure of Invention
In view of the above-mentioned deficiencies in the prior art, the present disclosure provides an intelligent question answering method, apparatus, computer device and computer readable medium.
In a first aspect, an embodiment of the present disclosure provides an intelligent question answering method, where the method includes:
acquiring an external request;
if the external request is an inquiry request, determining a conversation state, and performing semantic analysis on the inquiry request by using a preset semantic template to obtain semantic information;
if the inquiry request relates to a preset field and the conversation state is a non-interruption state, respectively inquiring a knowledge graph and an inquiry and answer engine according to the semantic information;
and generating answers according to the query results of the knowledge graph and the query results of the question answering engine.
In some embodiments, the semantic template is generated based on the knowledge-graph; the semantic analysis is performed on the inquiry request by using a preset semantic template to obtain semantic information, and the semantic information comprises the following steps:
and matching the inquiry request with a preset semantic template, and if the semantic template is hit, determining the semantic information of the hit semantic template.
In some embodiments, before semantically parsing the query request using a preset semantic template, the method further comprises: preprocessing the inquiry request;
the matching the query request with a preset semantic template includes:
and matching the preprocessed inquiry request with a preset semantic template.
In some embodiments, the preprocessing the query request includes one or any combination of the following:
merging the foreign words in the query request into foreign phrases;
removing symbols in the query request that do not contribute to semantic understanding;
and extracting a preset format text in the inquiry request.
In some embodiments, the performing semantic parsing on the query request by using a preset semantic template to obtain semantic information further includes:
if the semantic template is not hit and the query request relates to a preset field, extracting entities, attributes and relations from the query request, and extracting keyword strings;
determining a standard sentence corresponding to the inquiry request according to the entity, the attribute, the relationship, the keyword word string and a standard sentence inquiry library;
and determining semantic information of the standard sentence.
In some embodiments, said querying a question-answering engine according to said semantic information comprises:
querying a cache according to the semantic information;
if the corresponding query result is not queried in the cache, determining corresponding index information from the index database;
inquiring in a question-answering engine according to the index information to obtain an inquiry result of the question-answering engine;
generating a first query result set according to the query result of the question answering engine;
and storing the first query result set to a cache.
In some embodiments, said querying said knowledge-graph according to said semantic information comprises:
querying a cache according to the semantic information;
if the corresponding query result is not queried in the cache, calling a knowledge graph interface according to the semantic information, and performing knowledge graph query and relationship inference in a knowledge graph to obtain a query result of the knowledge graph;
generating a second query result set according to the query result of the knowledge graph;
and storing the second query result set to a cache.
In some embodiments, the generating answers from the query results of the knowledge-graph and the query results of the question-answering engine includes:
sequencing, screening, fusing and verifying the query result of the knowledge graph and the query result of the question-answering engine;
and generating answers described in natural language according to the query results of the knowledge graph and the query results of the question-answering engine after sequencing, screening, fusing and verifying.
In some embodiments, the intelligent question-answering method further comprises:
receiving a maintenance operation instruction, wherein the maintenance operation instruction comprises a knowledge graph maintenance operation instruction and/or a question-answering engine maintenance operation instruction;
executing the maintenance operation instruction, and recording the maintenance operation in a corresponding maintenance log;
and if an audit result that the maintenance operation in the maintenance log is not approved is received, rolling back the maintenance operation, and executing a maintenance operation instruction after the maintenance operation.
In some embodiments, after recording the maintenance operation in the corresponding maintenance log, further comprising:
in the maintenance log, marking the state of the maintenance operation record as a state to be audited;
the method further comprises the following steps:
and if an audit result that the maintenance operation in the maintenance log passes the audit is received, marking the state of the maintenance operation record that passes the audit as an audited state in the maintenance log.
On the other hand, the embodiment of the present disclosure further provides an intelligent question answering device, including: the system comprises an acquisition module, a processing module, a semantic analysis module, a query module and an answer generation module;
the acquisition module is used for acquiring an external request;
the processing module is used for determining a conversation state when the external request is an inquiry request;
the semantic analysis module is used for performing semantic analysis on the inquiry request by using a preset semantic template to obtain semantic information;
the query module is used for respectively querying a knowledge graph and a question-answering engine according to the semantic information when the query request relates to a preset field and the conversation state is a non-interruption state;
and the answer generating module is used for generating answers according to the query result of the knowledge graph and the query result of the question-answering engine.
In another aspect, an embodiment of the present disclosure further provides a computer device, including: one or more processors and storage; the storage device stores one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the intelligent question answering method provided in the foregoing embodiments.
The disclosed embodiments also provide a computer readable medium, on which a computer program is stored, wherein the computer program, when executed, implements the intelligent question answering method provided by the foregoing embodiments.
The intelligent question answering method provided by the embodiment of the disclosure comprises the following steps: acquiring an external request, if the external request is an inquiry request, determining a conversation state, and performing semantic analysis on the inquiry request by using a preset semantic template to obtain semantic information; if the inquiry request relates to a preset field and the conversation state is a non-interruption state, respectively inquiring a knowledge graph and a question-answering engine according to semantic information, and generating an answer according to the inquiry result of the knowledge graph and the inquiry result of the question-answering engine; the embodiment of the disclosure organically combines information retrieval based on unstructured corpora and knowledge maps based on structured knowledge, namely combines full-text retrieval and knowledge maps, and realizes intelligent question answering based on large-scale corpora and high-quality knowledge. In addition, on one hand, the structured knowledge provided by the knowledge map is introduced into the search question and answer, so that more professional and accurate answers can be provided, the traceability and interpretability of the answers can be enhanced, background information is provided for knowledge recommendation, a semantic template matching mechanism can accumulate semantic templates during system operation, and the system performance is improved in an incremental manner, so that the self-learning of the database is realized, and the workload of database operation and maintenance personnel is reduced; on the other hand, the intelligent question answering is introduced into the field of the knowledge graph, so that the accessibility of the knowledge graph can be greatly improved, and the value of the service knowledge accumulated for a long time can be fully exerted. The embodiment of the disclosure can be applied to intelligent customer service, knowledge management and intelligent marketing, reduces the burden of agent customer service personnel and background operation and maintenance management personnel, improves the service quality and knowledge management level, and improves the marketing efficiency.
Drawings
Fig. 1 is a flowchart of an intelligent question answering method provided in an embodiment of the present disclosure;
FIG. 2 is a flow chart of semantic parsing provided by embodiments of the present disclosure;
FIG. 3 is a flow chart of processing a query request provided by an embodiment of the present disclosure;
FIG. 4 is a flow diagram of a query answering engine provided by an embodiment of the present disclosure;
fig. 5 is a schematic structural diagram of a query unit of a question and answer engine provided in the embodiment of the present disclosure;
FIG. 6 is a flow diagram of querying a knowledge-graph provided by an embodiment of the present disclosure;
FIG. 7 is a schematic structural diagram of a knowledge-graph query unit provided in an embodiment of the present disclosure;
FIG. 8 is a schematic flow chart of database maintenance provided by an embodiment of the present disclosure;
fig. 9 is a schematic structural diagram of an intelligent question answering device according to an embodiment of the present disclosure.
Detailed Description
Example embodiments will be described more fully hereinafter with reference to the accompanying drawings, but which may be embodied in different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, 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.
Embodiments described herein may be described with reference to plan and/or cross-sectional views in light of idealized schematic illustrations of the disclosure. Accordingly, the example illustrations can be modified in accordance with manufacturing techniques and/or tolerances. Accordingly, the embodiments are not limited to the embodiments shown in the drawings, but include modifications of configurations formed based on a manufacturing process. Thus, the regions illustrated in the figures have schematic properties, and the shapes of the regions shown in the figures illustrate specific shapes of regions of elements, but are not intended to be limiting.
Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and the present disclosure, and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
Statistics shows that most of customer questions are concentrated on service details, and since service information behind the part of the questions is semi-structured or completely structured, the method is very suitable for introducing a knowledge graph to store, maintain, check, reason and recommend service knowledge. The knowledge graph is an extension of the semantic web technology without being tied to the semantic web. Compared with unstructured question-answer speech, the knowledge graph takes 'entities' as basic constituent units, and establishes semantic networks between the entities through 'relations' or 'attributes' between the entities, so that semantic knowledge retrieval and reasoning are achieved. A knowledge graph is a database that represents relationships between different entities and attributes of the entities. In the knowledge graph, entities are taken as nodes; the entities are connected with each other through edges, and the entities are connected with the values (attribute-value) of the attributes corresponding to the entities through edges, so that the structured and network-shaped database is formed.
One embodiment of the present disclosure provides an intelligent question-answering method, which extracts foreground information (i.e., a user request) and background environment (i.e., user information) of a user by using structured knowledge provided by a knowledge graph, performs professional knowledge reasoning and related knowledge recommendation by using the knowledge graph, provides interpretable traceable knowledge, processes interaction and feedback in user conversations, and reduces the operation and maintenance cost of a question-answering system.
The technology related to the embodiment of the disclosure mainly comprises:
(1) the semantic understanding scheme based on the telecommunication domain knowledge graph relates to the technologies including but not limited to: text mining, pattern matching, entity extraction, relationship extraction, intention extraction, text classification, ontology engineering and knowledge fusion.
(2) Knowledge retrieval, reasoning and recommendation schemes based on knowledge graph relate to technologies including but not limited to: the method comprises the steps of designing a special language in the field of graph query, designing and constructing an inference machine, and recommending knowledge based on user behaviors.
(3) The process of fusing knowledge graph and search-type question answering relates to the technology including but not limited to: knowledge map index, entity correlation algorithm and question-answer process arrangement.
(4) The scheme of dialog management by using knowledge graph includes but is not limited to: semantic slot filling, dialog state machine design based on user behavior and context information.
(5) The domain knowledge maintenance scheme based on the knowledge graph relates to the technologies including but not limited to: the method comprises the steps of knowledge graph operation transactionalization, knowledge verification, graph incremental maintenance, caching and auditing mechanisms.
The intelligent question-answering method is applied to an online operation scene, wherein the online operation scene refers to a scene of normally providing services to the outside, and answers can be provided for question and inquiry intelligence of a user in the scene. The on-line operation flow of the embodiment of the present disclosure is described in detail below with reference to fig. 1. As shown in fig. 1, the intelligent question answering method includes the following steps:
step 11, obtaining an external request.
In this step, the intelligent question answering device obtains an external request sent by the user, where the external request may be an inquiry request or an experience feedback request initiated by the user.
And step 12, if the external request is an inquiry request, determining the conversation state.
In this step, if the intelligent question answering device determines that the external request is a query request, the access control module is used to authenticate the query request, and after the authentication is passed, the dialog state is created or updated by matching a preset voice template. Dialog states may include an interrupted state and an uninterrupted state, the interrupted state indicating that a dialog is interrupted, e.g., a user saying "go to manual service"; if the non-interrupted state indicates that the dialogue is continued, the subsequent step of semantic parsing is performed on the inquiry request.
And step 13, performing semantic analysis on the inquiry request by using a preset semantic template to obtain semantic information.
In the embodiment of the present disclosure, the semantic information is a generalized concept, and may include semantics, an intention, keywords, a keyword group, a query statement, and the like. The specific process of performing semantic parsing on the query request by using the pre-established semantic template is described in detail with reference to fig. 2.
And 14, if the inquiry request relates to a preset field and the conversation state is a non-interruption state, respectively inquiring the knowledge graph and the inquiry and answer engine according to the semantic information.
In the embodiment of the disclosure, the preset field refers to the telecommunication field, and the telecommunication field has the characteristics of more users and more knowledge, so that the intelligent question-answering system is constructed in the telecommunication field, and the rapidly-growing customer service demand can be processed.
And respectively inquiring the specific processes of the knowledge graph and the question-answering engine according to the semantic information, and then explaining in detail.
And step 15, generating answers according to the query results of the knowledge graph and the query results of the question answering engine.
In this step, the generated answer is an answer described in natural language so as to be easily understood by the user, and the answer may be returned to the user together with information to be fed back by the user.
In some embodiments, the query results of the knowledge graph and the query results of the question and answer engine are ranked, screened, fused and verified, and answers described in natural language are generated according to the query results of the knowledge graph and the query results of the question and answer engine after the query results of the knowledge graph and the query results of the question and answer engine are processed through ranking, screening, fusing and verifying.
As can be seen from steps 11 to 15, the intelligent question answering method provided by the embodiment of the present disclosure includes: acquiring an external request, if the external request is an inquiry request, determining a conversation state, and performing semantic analysis on the inquiry request by using a preset semantic template to obtain semantic information; if the inquiry request relates to a preset field and the conversation state is a non-interruption state, respectively inquiring a knowledge graph and a question-answering engine according to semantic information, and generating an answer according to the inquiry result of the knowledge graph and the inquiry result of the question-answering engine; the embodiment of the disclosure organically combines information retrieval based on unstructured corpora and knowledge maps based on structured knowledge, namely combines full-text retrieval and knowledge maps, and realizes intelligent question answering based on large-scale corpora and high-quality knowledge. In addition, on one hand, the structured knowledge provided by the knowledge map is introduced into the search question and answer, so that more professional and accurate answers can be provided, the traceability and interpretability of the answers can be enhanced, background information is provided for knowledge recommendation, a semantic template matching mechanism can accumulate semantic templates during system operation, and the system performance is improved in an incremental manner, so that the self-learning of the database is realized, and the workload of database operation and maintenance personnel is reduced; on the other hand, the intelligent question answering is introduced into the field of the knowledge graph, so that the accessibility of the knowledge graph can be greatly improved, and the value of the service knowledge accumulated for a long time can be fully exerted. The embodiment of the disclosure can be applied to intelligent customer service, knowledge management and intelligent marketing, reduces the burden of agent customer service personnel and background operation and maintenance management personnel, improves the service quality and knowledge management level, and improves the marketing efficiency.
In some embodiments, in step 12, if the external request is an experience feedback type request, the dialog state is determined, and the existing data in the knowledge graph and the question-answering engine are labeled and cached according to the user feedback and the dialog state information.
In some embodiments, the semantic template is generated based on a knowledge graph. The following describes the semantic parsing process in detail with reference to fig. 2. As shown in fig. 2, the semantic parsing (i.e. step 13) of the query request by using the pre-established semantic template includes the following steps:
step 131, matching the query request with the semantic template, and if the query request hits the semantic template, executing step 137; if the query request does not hit the semantic template, step 132 is performed.
In the step, the inquiry request is matched with a pre-stored semantic template, and meanwhile, the generalization capability of template matching is realized by combining semantic associated information provided by a knowledge graph. The semantic template is not a simple character string pattern or a regular expression, but a statement template containing keyword arguments, for example, the preset semantic template is "$ { E: package } how to charge", wherein "$ { E: package }" represents all entities belonging to the package entity in the knowledge graph, so that the semantic template can be matched with a large class of inquiry requests of all "XX package how to charge", and preliminary structuring of natural language question sentences can be realized. It should be noted that, the embodiment of the present disclosure does not limit the specific format of the semantic template, and the above is only one example of the semantic template.
Step 132, determining the field of the inquiry request, and determining whether the field relates to a preset field, if so, step 133, otherwise, returning the inquiry request.
When the inquiry request is matched with the semantic template, in the step, the inquiry request is subjected to domain classification by using a domain classification algorithm, and the related domain is determined, wherein the domain classification algorithm is pluggable, can be configured on line and is trained on line. It should be noted that the embodiments of the present disclosure do not limit the specific algorithm for determining the domain to which the query request relates. If the intelligent question answering device judges that the inquiry request is in the telecommunication field, information extraction is carried out (namely, step 133 is executed); if the inquiry request is judged to be in the non-telecommunication domain, the inquiry request and the result generated in the step 131 are directly returned.
Step 133 extracts entities, attributes and relationships for the query request.
In this step, the intelligent question answering device extracts key information such as entities, attributes, relationships and the like appearing in the inquiry request by comprehensively using algorithms such as rule matching, word stock matching, machine learning, deep learning and the like, and fills the extracted information into slot combinations designed according to specific services to complete basic question sentence structuring.
Step 134, extracting keyword strings.
In this step, the intelligent question-answering device extracts keyword strings from the short text of the inquiry request by comprehensively using a plurality of algorithms, and selects the first N keyword strings according to the sentence length. The keywords are clustered into two aspects: the keyword string with higher semantic importance calculated by using the algorithm is used on one hand, and the keyword string matched with the preset keyword library is used on the other hand.
And step 135, determining a standard sentence corresponding to the inquiry request according to the entity, the attribute, the relation, the keyword word string and the standard inquiry sentence library.
In this step, similarity matching is performed on the semantics of the query request and the existing standard sentences in the standard question library, so that the query request is standardized into the existing standard sentences in the system, and the purpose of equivalent sentence normalization is to extract information. It should be noted that the embodiment of the present disclosure does not limit the similarity calculation and the equivalent sentence matching algorithm.
At step 136, semantic information of the standard sentence is determined.
Each standard sentence may be considered to correspond to a particular intent, and therefore equivalence sentence normalization may enable intent recognition.
In step 137, semantic information of the hit semantic template is determined.
In this step, when the query request hits a semantic template, the semantic information of the hit semantic template can be directly determined.
Through the steps 131 to 137, the semantic template matching mechanism can deeply couple the semantic information provided by the knowledge graph and the symbolic structure information of the semantic template, innovatively realize understanding and reasoning of a knowledge level in a semantic understanding stage of a question, and improve the accuracy and coverage rate of answer retrieval. Meanwhile, the semantic template matching mechanism is low in complexity, does not relate to the transformation of hardware equipment, can completely realize incremental updating, and even can realize hot updating when the intelligent question-answering device runs. On the other hand, the semantic template matching mechanism is easy to master for operation and maintenance personnel, the learning curve is gentle, richer semantic templates can be accumulated along with the operation of the intelligent question answering device, the system performance is improved in an incremental mode, and then the manpower maintenance cost is reduced gradually. The method has great significance for telecom operators, can reduce the modification cost and the downtime to the maximum extent, and improves the cost-effectiveness ratio.
In some embodiments, as shown in fig. 3, before semantically parsing the query request using the preset semantic template (i.e., step 13), the method further comprises the steps of:
step 13', the inquiry request is preprocessed.
In some embodiments, the query request is preprocessed, which may include, but is not limited to, one or any combination of the following: segmenting the query request according to a segmentation word bank associated with the knowledge graph, and removing stop words; merging the foreign words in the query request into foreign phrases; removing symbols in the query request that do not contribute to semantic understanding; and extracting a format text preset in the inquiry request.
In this step, the intelligent question-answering device performs preliminary word segmentation and stop word processing on the inquiry request by using a word segmentation word bank associated with the knowledge map, wherein the word segmentation word bank is a vocabulary set provided by the knowledge map. The implementation of the used word segmentation dictionary, the word weight, the stop word dictionary and even the word segmentation algorithm is configurable and pluggable, so that the vocabulary information provided by the map can be fully utilized, and the knowledge map word segmentation lexicon can be independently expanded according to the requirement. The foreign language phrases and phrases after being divided into words and removed from stop words can be combined into complete foreign language phrases according to the information provided by the word stock. Special symbols which do not contribute to semantic understanding are removed, but key symbols such as question marks, decimal points and the like are not removed. And extracting special format texts such as time, date, province names, mobile phone numbers, flow numbers and the like so as to prepare for semantic understanding of information such as numbers and the like.
If the preprocessing step is performed, correspondingly, the matching the query request with the preset semantic template (i.e. step 131) includes: and matching the preprocessed inquiry request with a preset semantic template.
The flow of the query answering engine is described in detail below with reference to fig. 4 and 5.
In some embodiments, as shown in fig. 4 and 5, the querying the question-answering engine according to semantic information includes the following steps:
step 21, inquiring cache according to semantic information.
The cache is used for recording query history and query results, and comprises a cache which is used for recording and counting high-frequency queries so as to improve query efficiency.
And step 22, if the corresponding query result is not queried in the cache, determining corresponding index information from the index database.
The index database is used for storing index content, and in the step, if the intelligent question answering device does not inquire the corresponding inquiry result in the cache, the index information corresponding to the semantic information is determined from the index database so as to be inquired in the question answering engine based on the index information.
And step 23, inquiring in the question answering engine according to the index information to obtain an inquiry result of the question answering engine.
The question-answer engine access interface is used for uniformly managing access operation on a question-answer full-text database, and the question-answer full-text database is used for storing unstructured question-answer corpora. In this step, the intelligent question-answering device queries the question-answering engine through the question-answering engine access interface to obtain a query result corresponding to the index information.
And 24, generating a first query result set according to the query result of the question answering engine.
In this step, the query results of the queried question-answering engine are ranked according to the relevancy algorithm to generate a first query result set. The relevancy ranking algorithm is pluggable and configurable.
Step 25, storing the first query result set to a cache.
It should be noted that, when maintaining the question-answering engine, an indexer is used to build an index for the full-text corpus (i.e., the question-answering engine) by using the information such as the knowledge map word segmentation lexicon and the relationships between entities, and the built index is stored in the index repository.
The answer which is most matched with the query request can be retrieved from a large-scale corpus of the whole texts through the steps 21-25, and a high recall ratio is ensured. In order to ensure the indexing effect and efficiency, the high-quality knowledge provided by the knowledge graph can be maximally utilized by utilizing the word segmentation word stock provided by the knowledge graph module to construct the index.
The flow of the query answering engine is described in detail below with reference to fig. 6 and 7.
In some embodiments, as shown in fig. 6 and 7 in combination, the querying the knowledge-graph according to the semantic information includes the following steps:
step 31, querying the cache according to the semantic information.
In this step, the query parser parses the query statement (i.e. structured query command) in the semantic information, and configures a file by using the format of the query command, so as to unify the specific format of the semantics and the intention. The query parser is used as an external interface of the knowledge graph query unit, can compile query statements into query operation instructions which can be directly executed, coordinates and schedules each sub-module of the knowledge graph query unit to work, ensures that the query of an external module on the knowledge graph meets the transaction characteristics, and simultaneously optimizes the query performance according to a certain optimization rule.
The query log and the result cache can record query and maintenance records of the knowledge graph for query analysis and graph management, and can also cache high-frequency query to improve retrieval performance.
And step 32, if the corresponding query result is not queried in the cache, calling a knowledge graph interface according to the semantic information, and performing knowledge graph query and relationship inference in the knowledge graph to obtain a query result of the knowledge graph.
The knowledge graph interface is used for providing an addition, deletion, modification and query interface of the knowledge graph and isolating the knowledge graph. The knowledge graph is used for persistently storing the structural knowledge in the knowledge graph, and any database scheme which can be used for knowledge storage can be adopted, including RDF (Resource Description Framework), a graph database, a document database, a relational database and the like.
The inference machine is used for executing knowledge graph query and relation inference, and querying corresponding structured knowledge, such as entity ID, answer ID, external Resource URI (Uniform Resource Identifier) or triple statement, from the knowledge graph by comprehensively adopting multiple algorithms such as graph retrieval, graph matching, type derivation, graph embedding operation and the like based on the knowledge of a domain ontology, associated knowledge, a semantic template, an inference rule and the like defined or stored in the knowledge graph according to a query statement provided by the query analyzer.
And step 33, generating a second query result set according to the query result of the knowledge graph.
In the step, according to the checking rule, strategy and/or algorithm, the inquired knowledge graph inquiry results are sorted, screened, fused and verified, and a second inquiry result set is constructed. The purpose of verification is to ensure that the retrieval result of the knowledge graph module is correct and no contradiction or even error occurs.
Step 34, storing the second query result set to a cache.
Through the steps 31-34, the required structural information can be efficiently and accurately retrieved from the knowledge graph according to the query statement, and the precision ratio of the system and the credibility and interpretability of the query result are greatly improved. Meanwhile, the process can be regarded as encapsulation of knowledge graph storage, so that all operations on the knowledge graph meet the transaction characteristics of atomicity, consistency, isolation and durability.
In some embodiments, the intelligent question-answering method may also be applied to an offline maintenance scenario, where the offline maintenance refers to that in the morning or other periods, external services are temporarily stopped, and maintenance operations such as data modification, data update, data import and export, system test, troubleshooting, and the like are performed on a system by a maintenance worker. The most important requirement of the offline maintenance scenario is the update and modification of data in the database. In the disclosed embodiments, the data includes, but is not limited to, structured knowledge stored in a knowledge graph, unstructured or semi-structured full-text information stored in a question and answer engine, inference rules, templates, flow scripts, configuration files, metadata, and the like.
The offline operation flow of the embodiment of the present disclosure is described in detail below with reference to fig. 8. As shown in fig. 8, the intelligent question answering method further includes the following steps:
and step 41, receiving a maintenance operation instruction, wherein the maintenance operation instruction comprises a knowledge graph maintenance operation instruction and/or a question-answering engine maintenance operation instruction.
The maintenance personnel of the database are divided into two types according to the authority size, namely a common manager and a high-level manager, the maintenance operation instruction can be sent by the common manager, and the daily maintenance operation can comprise adding a triple and deleting the question and answer which are too late.
In this step, the library maintenance operation command is authenticated and then cached and sent to the maintenance module of the intelligent question-answering device.
And step 42, executing the maintenance operation instruction and recording the maintenance operation in a corresponding maintenance log.
The intelligent question answering device executes the maintenance operation instruction and simultaneously records the maintenance log for record. The quality of the database content is of vital importance and may even affect the operation of the actual service. Because the manual maintenance of the database inevitably causes errors, the intelligent question answering device records a specific maintenance log when executing a maintenance operation instruction and modifying the content of the database. The content of the maintenance log includes, but is not limited to, the state of the database before operation, the state of the database after operation, the operator, the operation time, and other key information. The content setting of the maintenance log meets the traceable characteristic, namely the complete maintenance log and the state of the maintained database are known, and the database can be completely recovered to the state before the maintenance operation instruction is executed by reversely pushing the maintenance log.
And 43, rolling back the maintenance operation if an audit result that the maintenance operation in the maintenance log is not approved is received.
In the step, the high-level administrator checks the maintenance logs one by one according to the time sequence, if an incorrect maintenance log record is found, which indicates that the log record is not checked, a rollback flow is started, and according to the maintenance log and the state of the database, the state of the database is rolled back to the state before the maintenance operation instruction is executed, so as to eliminate the influence of the error operation on the database.
In order to ensure the consistency of the data of the knowledge base, the maintenance operation can be checked and passed only if all the maintenance operations before a certain maintenance operation are checked and passed. If a maintenance operation does not pass the audit, the operation must be rolled back. The rollback operation does not affect the previous maintenance operation, but the subsequent maintenance operation needs the audit by the manager one by one to ensure that no conflict occurs.
And step 44, executing the maintenance operation command after the maintenance operation.
In this step, since the previous maintenance operation (i.e., the error operation) is rolled back and the subsequent correct operation does not exist, the maintenance operation (i.e., the correct operation) instruction after the error operation needs to be executed again to ensure that the state of the database is up to date.
In some embodiments, after recording the maintenance operation in the corresponding maintenance log (i.e., step 42), the intelligent question-answering method further comprises: and in the maintenance log, marking the state of the maintenance operation record as a state to be audited for high-level administrators to audit. That is, after the maintenance operation is written into the cache, a maintenance record marked as "to be audited" is written into the maintenance log, and the maintenance operation does not actually act on the database at this time.
Correspondingly, the intelligent question answering method further comprises the following steps: and if an audit result that the maintenance operation in the maintenance log passes the audit is received, marking the state of the maintenance operation record passing the audit as an audited state in the maintenance log.
In the field of telecommunications, telecommunication operation business is complicated, background operation and maintenance management and customer service seat personnel are numerous, data confusion (such as misoperation, data which is too old or wrong, data conflict under the condition of multi-person cooperation and the like) in a database is easily caused in the daily operation process, the order degree of the knowledge graph is further damaged in a linkage mode, the intelligence of the knowledge graph is lost, even the knowledge graph is missed, and the knowledge graph falls back to a traditional knowledge management system without semantics or with slight semantics. The embodiment of the disclosure innovatively brings human factors into the consideration scope of the intelligent question-answering scheme again, and pays attention to the knowledge mining from the organization and the mind of the human, and meanwhile, technical and management means are used for ensuring that human intervention does not introduce extra entropy increase to the system, so that the cleanness and the order of the system are ensured, and the method has great and general significance to intelligent systems in the field and all similar fields.
The method and the system can combine the traditional retrieval type question-answering system and the knowledge map technology, the question-answering part adopts a retrieval type scheme, and the knowledge map retrieval process is innovatively embedded into the processing process of the question-answering system through methods such as entity matching, slot filling, template matching and translation. In order to adapt retrieval, one side of the knowledge graph provides a corresponding query and access interface, so that the interoperation of the query of a question and answer engine and the query of the knowledge graph can be realized, and the high cohesion of each database can be ensured. The core of the retrieval type question-answering is a highly optimized inverted index and relevance calculation algorithm, and special hardware and software environment support is not needed; the knowledge graph storage is based on a relational database which is widely used in the telecommunication industry and mature in operation and maintenance technology, so that the intelligent question-answering method provided by the embodiment of the disclosure has practicability.
Based on the same technical concept, an embodiment of the present disclosure further provides an intelligent question answering device, as shown in fig. 9, the intelligent question answering device includes: the system comprises an acquisition module 1, a processing module 2, a semantic analysis module 3, an inquiry module 4 and an answer generation module 5, wherein the acquisition module 1 is used for acquiring an external request.
The processing module 2 is configured to determine a session state when the external request is an inquiry request.
The semantic analysis module 3 is used for performing semantic analysis on the inquiry request by using a preset semantic template to obtain semantic information.
And the query module 4 is used for respectively querying the knowledge graph and the question-answering engine according to the semantic information when the query request relates to a preset field and the conversation state is a non-interruption state.
And the answer generating module 5 is used for generating answers according to the query result of the knowledge graph and the query result of the question-answering engine.
In some embodiments, semantic templates are generated based on the knowledge-graph.
And the semantic analysis module 3 is used for matching the inquiry request with a preset semantic template, and determining the semantic information of the hit semantic template when the semantic template is hit.
In some embodiments, the intelligent question-answering device further includes a preprocessing module, and the preprocessing module is configured to preprocess the query request before the semantic parsing module 3 performs semantic parsing on the query request by using a preset semantic template.
Correspondingly, the semantic parsing module 3 is configured to match the preprocessed inquiry request with a preset semantic template.
In some embodiments, the pre-processing of the query request includes, but is not limited to, one or any combination of the following:
segmenting the query request according to a segmentation word bank associated with the knowledge graph, and removing stop words;
merging the foreign words in the query request into foreign phrases;
removing symbols in the query request that do not contribute to semantic understanding;
and extracting a preset format text in the inquiry request.
In some embodiments, the semantic parsing module 3 is further configured to, when the semantic template is not hit and the query request relates to a preset domain, extract an entity, an attribute, and a relationship for the query request, and extract a keyword string; determining a standard sentence corresponding to the inquiry request according to the entity, the attribute, the relationship, the keyword word string and a standard sentence inquiry library; and determining semantic information of the standard sentence.
In some embodiments, the query module 4 comprises a question-and-answer engine query unit, configured to query the cache according to the semantic information; when the corresponding query result is not queried in the cache, determining corresponding index information from the index library; inquiring in a question-answering engine according to the index information to obtain an inquiry result of the question-answering engine; generating a first query result set according to the query result of the question answering engine; and storing the first query result set to a cache.
In some embodiments, the query module 4 comprises a knowledge-graph query unit configured to query the cache according to the semantic information; if the corresponding query result is not queried in the cache, calling a knowledge graph interface according to the semantic information, and performing knowledge graph query and relationship inference in a knowledge graph to obtain a query result of the knowledge graph;
generating a second query result set according to the query result of the knowledge graph; and storing the second query result set to a cache.
In some embodiments, the answer generation module 5 is configured to rank, filter, fuse, and verify the query results of the knowledge graph and the query results of the question-answering engine; and generating answers described in natural language according to the query results of the knowledge graph and the query results of the question-answering engine after sequencing, screening, fusing and verifying.
The intelligent question-answering device also comprises a maintenance module, wherein the maintenance module is used for receiving a maintenance operation instruction, and the maintenance operation instruction comprises a knowledge graph maintenance operation instruction and/or a question-answering engine maintenance operation instruction; executing the maintenance operation instruction, and recording the maintenance operation in a corresponding maintenance log; and when an auditing result that the maintenance operation in the maintenance log is not approved is received, rolling back the maintenance operation, and executing a maintenance operation instruction after the maintenance operation.
In some embodiments, the maintenance module is further configured to, after recording the maintenance operation in the corresponding maintenance log, mark the status of the maintenance operation record in the maintenance log as a to-be-audited status; when an auditing result that the auditing of the maintenance operation in the maintenance log is passed is received, the state of the record of the maintenance operation that the auditing is passed is marked as an audited state in the maintenance log.
An embodiment of the present disclosure further provides a computer device, including: one or more processors and storage; the storage device stores one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the intelligent question answering method provided in the foregoing embodiments.
The disclosed embodiments also provide a computer readable medium, on which a computer program is stored, wherein the computer program, when executed, implements the intelligent question answering method provided by the foregoing embodiments.
It will be understood by those of ordinary skill in the art that all or some of the steps of the methods disclosed above, functional modules/units in the apparatus, may be implemented as software, firmware, hardware, and suitable combinations thereof. In a hardware implementation, the division between functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be performed by several physical components in cooperation. Some or all of the physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as is well known to those of ordinary skill in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can accessed by a computer. In addition, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media as known to those skilled in the art.
Example embodiments have been disclosed herein, and although specific terms are employed, they are used and should be interpreted in a generic and descriptive sense only and not for purposes of limitation. In some instances, features, characteristics and/or elements described in connection with a particular embodiment may be used alone or in combination with features, characteristics and/or elements described in connection with other embodiments, unless expressly stated otherwise, as would be apparent to one skilled in the art. It will, therefore, be understood by those skilled in the art that various changes in form and details may be made therein without departing from the scope of the invention encompassed by the appended claims.
Claims (13)
1. A method of intelligent question answering, the method comprising:
acquiring an external request;
if the external request is an inquiry request, determining a conversation state, and performing semantic analysis on the inquiry request by using a preset semantic template to obtain semantic information;
if the inquiry request relates to a preset field and the conversation state is a non-interruption state, respectively inquiring a knowledge graph and an inquiry and answer engine according to the semantic information;
and generating answers according to the query results of the knowledge graph and the query results of the question answering engine.
2. The method of claim 1, wherein the semantic template is generated based on the knowledge-graph; the semantic analysis is performed on the inquiry request by using a preset semantic template to obtain semantic information, and the semantic information comprises the following steps:
and matching the inquiry request with a preset semantic template, and if the semantic template is hit, determining the semantic information of the hit semantic template.
3. The method of claim 2, wherein prior to semantically parsing the query request using a preset semantic template, the method further comprises: preprocessing the inquiry request;
the matching the query request with a preset semantic template includes:
and matching the preprocessed inquiry request with a preset semantic template.
4. The method of claim 3, wherein the preprocessing the query request comprises one or any combination of the following:
segmenting the query request according to a segmentation word bank associated with the knowledge graph, and removing stop words;
merging the foreign words in the query request into foreign phrases;
removing symbols in the query request that do not contribute to semantic understanding;
and extracting a preset format text in the inquiry request.
5. The method of claim 2, wherein the parsing the query request using a predetermined semantic template to obtain semantic information further comprises:
if the semantic template is not hit and the query request relates to a preset field, extracting entities, attributes and relations from the query request, and extracting keyword strings;
determining a standard sentence corresponding to the inquiry request according to the entity, the attribute, the relationship, the keyword word string and a standard sentence inquiry library;
and determining semantic information of the standard sentence.
6. The method of claim 1, wherein said querying a question and answer engine based on said semantic information comprises:
querying a cache according to the semantic information;
if the corresponding query result is not queried in the cache, determining corresponding index information from the index database;
inquiring in a question-answering engine according to the index information to obtain an inquiry result of the question-answering engine;
generating a first query result set according to the query result of the question answering engine;
and storing the first query result set to a cache.
7. The method of claim 1, wherein the querying the knowledge-graph according to the semantic information comprises:
querying a cache according to the semantic information;
if the corresponding query result is not queried in the cache, calling a knowledge graph interface according to the semantic information, and performing knowledge graph query and relationship inference in a knowledge graph to obtain a query result of the knowledge graph;
generating a second query result set according to the query result of the knowledge graph;
and storing the second query result set to a cache.
8. The method of claim 1, wherein the generating answers from the query results of the knowledge-graph and the query results of the question-answering engine comprises:
sequencing, screening, fusing and verifying the query result of the knowledge graph and the query result of the question-answering engine;
and generating answers described in natural language according to the query results of the knowledge graph and the query results of the question-answering engine after sequencing, screening, fusing and verifying.
9. The method of any one of claims 1-8, wherein the method further comprises:
receiving a maintenance operation instruction, wherein the maintenance operation instruction comprises a knowledge graph maintenance operation instruction and/or a question-answering engine maintenance operation instruction;
executing the maintenance operation instruction, and recording the maintenance operation in a corresponding maintenance log;
and if an audit result that the maintenance operation in the maintenance log is not approved is received, rolling back the maintenance operation, and executing a maintenance operation instruction after the maintenance operation.
10. The method of claim 9, wherein after recording the maintenance operations in the respective maintenance logs, further comprising:
in the maintenance log, marking the state of the maintenance operation record as a state to be audited;
the method further comprises the following steps:
and if an audit result that the maintenance operation in the maintenance log passes the audit is received, marking the state of the maintenance operation record that passes the audit as an audited state in the maintenance log.
11. An intelligent question answering device comprising: the system comprises an acquisition module, a processing module, a semantic analysis module, a query module and an answer generation module;
the acquisition module is used for acquiring an external request;
the processing module is used for determining a conversation state when the external request is an inquiry request;
the semantic analysis module is used for performing semantic analysis on the inquiry request by using a preset semantic template to obtain semantic information;
the query module is used for respectively querying a knowledge graph and a question-answering engine according to the semantic information when the query request relates to a preset field and the conversation state is a non-interruption state;
and the answer generating module is used for generating answers according to the query result of the knowledge graph and the query result of the question-answering engine.
12. A computer device, comprising:
one or more processors;
a storage device having one or more programs stored thereon;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the intelligent question answering method of any one of claims 1-10.
13. A computer-readable medium on which a computer program is stored, wherein the program, when executed, implements the intelligent question-answering method according to any one of claims 1 to 10.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911328449.XA CN113010632A (en) | 2019-12-20 | 2019-12-20 | Intelligent question answering method and device, computer equipment and computer readable medium |
| PCT/CN2020/113778 WO2021120707A1 (en) | 2019-12-20 | 2020-09-07 | Intelligent question-answering method and apparatus, computer device, and computer-readable medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911328449.XA CN113010632A (en) | 2019-12-20 | 2019-12-20 | Intelligent question answering method and device, computer equipment and computer readable medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113010632A true CN113010632A (en) | 2021-06-22 |
Family
ID=76382848
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911328449.XA Pending CN113010632A (en) | 2019-12-20 | 2019-12-20 | Intelligent question answering method and device, computer equipment and computer readable medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN113010632A (en) |
| WO (1) | WO2021120707A1 (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113590838A (en) * | 2021-07-30 | 2021-11-02 | 上海明略人工智能(集团)有限公司 | Customer service enabling method and system based on knowledge graph and storage medium |
| WO2023028780A1 (en) * | 2021-08-30 | 2023-03-09 | Siemens Aktiengesellschaft | Method, apparatus and system for graph data caching |
| CN116524926A (en) * | 2023-04-27 | 2023-08-01 | 百洋智能科技集团股份有限公司 | Method for generating service form through voice control at mobile terminal |
| CN116737964A (en) * | 2023-08-04 | 2023-09-12 | 联想新视界（北京）科技有限公司 | Artificial intelligence brain system |
| CN117130791A (en) * | 2023-10-26 | 2023-11-28 | 南通话时代信息科技有限公司 | Computing power resource allocation method and system of cloud customer service platform |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114579605B (en) * | 2022-04-26 | 2022-08-09 | 阿里巴巴达摩院(杭州)科技有限公司 | Table question-answer data processing method, electronic equipment and computer storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9842166B1 (en) * | 2014-08-08 | 2017-12-12 | Google Llc | Semi structured question answering system |
| CN105868313B (en) * | 2016-03-25 | 2019-02-12 | 浙江大学 | A kind of knowledge mapping question answering system and method based on template matching technique |
| CN109492077B (en) * | 2018-09-29 | 2020-09-29 | 北京智通云联科技有限公司 | Knowledge graph-based petrochemical field question-answering method and system |
| CN110147436B (en) * | 2019-03-18 | 2021-02-26 | 清华大学 | Education knowledge map and text-based hybrid automatic question-answering method |
| CN110399457B (en) * | 2019-07-01 | 2023-02-03 | 吉林大学 | Intelligent question answering method and system |
-
2019
- 2019-12-20 CN CN201911328449.XA patent/CN113010632A/en active Pending
-
2020
- 2020-09-07 WO PCT/CN2020/113778 patent/WO2021120707A1/en active Application Filing
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113590838A (en) * | 2021-07-30 | 2021-11-02 | 上海明略人工智能(集团)有限公司 | Customer service enabling method and system based on knowledge graph and storage medium |
| WO2023028780A1 (en) * | 2021-08-30 | 2023-03-09 | Siemens Aktiengesellschaft | Method, apparatus and system for graph data caching |
| CN116524926A (en) * | 2023-04-27 | 2023-08-01 | 百洋智能科技集团股份有限公司 | Method for generating service form through voice control at mobile terminal |
| CN116737964A (en) * | 2023-08-04 | 2023-09-12 | 联想新视界（北京）科技有限公司 | Artificial intelligence brain system |
| CN116737964B (en) * | 2023-08-04 | 2023-11-17 | 联想新视界（北京）科技有限公司 | Artificial intelligence brain system |
| CN117130791A (en) * | 2023-10-26 | 2023-11-28 | 南通话时代信息科技有限公司 | Computing power resource allocation method and system of cloud customer service platform |
| CN117130791B (en) * | 2023-10-26 | 2023-12-26 | 南通话时代信息科技有限公司 | Computing power resource allocation method and system of cloud customer service platform |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021120707A1 (en) | 2021-06-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11790006B2 (en) | Natural language question answering systems | |
| CN113010632A (en) | Intelligent question answering method and device, computer equipment and computer readable medium | |
| US11442932B2 (en) | Mapping natural language to queries using a query grammar | |
| US10896212B2 (en) | System and methods for automating trademark and service mark searches | |
| WO2020007224A1 (en) | Knowledge graph construction and smart response method and apparatus, device, and storage medium | |
| US8972408B1 (en) | Methods, systems, and articles of manufacture for addressing popular topics in a social sphere | |
| CN108922633A (en) | A kind of disease name standard convention method and canonical system | |
| US20140222822A1 (en) | Content revision using question and answer generation | |
| CN110765277B (en) | Knowledge-graph-based mobile terminal online equipment fault diagnosis method | |
| US11403286B2 (en) | Bridge from natural language processing engine to database engine | |
| US9606984B2 (en) | Unsupervised clustering of dialogs extracted from released application logs | |
| US11226960B2 (en) | Natural-language database interface with automated keyword mapping and join-path inferences | |
| Mariani et al. | Semantic matching of gui events for test reuse: are we there yet? | |
| CN112115232A (en) | Data error correction method and device and server | |
| CN110597844A (en) | Heterogeneous database data unified access method and related equipment | |
| CN114218472A (en) | Intelligent search system based on knowledge graph | |
| Cheng et al. | A similarity integration method based information retrieval and word embedding in bug localization | |
| CN112445813A (en) | Search semantic analysis method for enterprise service portal platform | |
| EP4295245A1 (en) | Methods and systems for controlled modeling and optimization of a natural language database interface | |
| CN111831624A (en) | Data table creating method and device, computer equipment and storage medium | |
| US20220269706A1 (en) | Methods and systems to parse a software component search query to enable multi entity search | |
| CN113392202A (en) | Knowledge graph-based question-answering system and method | |
| Lehmberg | Web table integration and profiling for knowledge base augmentation | |
| CN114391142A (en) | Parsing queries using structured and unstructured data | |
| US11860914B1 (en) | Natural language database generation and query system |
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 |