Search

CN-121979994-A - Intelligent question-answering method, device, equipment and medium

CN121979994ACN 121979994 ACN121979994 ACN 121979994ACN-121979994-A

Abstract

The invention relates to the technical field of artificial intelligence, which can be applied to the fields of financial science and technology and medical health, and discloses an intelligent question-answering method, device, equipment and medium, wherein the method comprises the steps of receiving natural language questions input by a user, and processing the natural language questions to obtain business entities and query intents; the method comprises the steps of matching and disambiguating the service entity with term nodes in a pre-constructed service term library to determine standard service terms corresponding to the service entity in the service term library, generating executable data query sentences based on the standard service terms, the query intention and a preset query template, executing the data query sentences to obtain data results, converting the data results into natural language replies and outputting the natural language replies. The accuracy of intelligent question and answer is improved.

Inventors

  • QU XIAOYANG

Assignees

  • 平安科技（深圳）有限公司

Dates

Publication Date
20260505
Application Date
20260202

Claims (10)

  1. 1. An intelligent question-answering method is characterized by comprising the following steps: Receiving a natural language question input by a user, and processing the natural language question to obtain a service entity and a query intention; matching and disambiguating the service entity with term nodes in a pre-constructed service term library to determine standard service terms corresponding to the service entity in the service term library; generating executable data query sentences based on the standard business terms, the query intents and preset query templates; executing the data query statement to obtain a data result, converting the data result into a natural language reply and outputting the natural language reply.
  2. 2. The intelligent question-answering method according to claim 1, wherein the step of processing the natural language question to obtain a business entity and a query intention comprises: Word segmentation is carried out on the natural language problem to obtain a word segmentation natural language problem; Filtering the word segmentation natural language problem to remove the virtual words and the stop words in the word segmentation natural language problem, so as to obtain a target natural language problem; And identifying and classifying the target natural language problem to obtain the business entity and the query intention.
  3. 3. The intelligent question-answering method according to claim 2, wherein the step of identifying and classifying the target natural language question to obtain the business entity and the query intention comprises: Identifying a service entity in the target natural language problem by adopting a named entity identification model, and labeling the service entity with a service exclusive type label; and carrying out intention classification processing on the target natural language problem by adopting an intention classification model so as to determine a predefined query intention category to which the target natural language problem belongs, and obtaining the query intention.
  4. 4. The intelligent question-answering method according to claim 3, wherein after the step of using a named entity recognition model to recognize a business entity in the target natural language question and labeling the business entity with a business-specific type tag, the method further comprises: If the service exclusive type label marked by the service entity is a time expression, then The temporal expression is normalized to a time range condition using a time resolution rule base.
  5. 5. The intelligent question-answering method according to claim 1, wherein the step of matching and disambiguating the business entity with term nodes in a pre-constructed business term library to determine standard business terms corresponding to the business entity in the business term library comprises: Searching candidate term nodes related to the business entity from the pre-constructed business term library based on semantic similarity; Based on the graph structure relationship in the business term library, analyzing the relevance between the candidate term nodes and the term nodes determined in the natural language questions; and determining the standard service term which is finally matched from the candidate term nodes according to the strength of the relevance.
  6. 6. The intelligent question-answering method according to claim 1, wherein the step of generating an executable data query sentence based on the standard business term, the query intention, and a preset query template, comprises: generating a query intermediate representation structure according to the standard business term and the query intention, wherein the query intermediate representation structure comprises indexes, dimensions, screening conditions and query type information; And rendering and generating the executable data query statement by using a template engine according to the query intermediate representation structure and the preset query template, and adding a data filtering condition into the query intermediate representation structure based on the access authority of a user in the process of assembling and generating the data query statement, or embedding an authority filtering clause when rendering and generating the data query statement.
  7. 7. The intelligent question-answering method according to any one of claims 1 to 6, wherein the business term library is constructed by using a graph database including term nodes representing different business concepts and relationship edges representing relationships between the term nodes, wherein the term nodes include index nodes and dimension nodes, and the relationship edges include mapping relationships.
  8. 8. An intelligent question-answering device, comprising: the receiving and processing unit is used for receiving natural language questions input by a user and processing the natural language questions to obtain service entities and query intents; The matching disambiguation unit is used for matching and disambiguating the service entity with term nodes in a pre-constructed service term library so as to determine standard service terms corresponding to the service entity in the service term library; the generation unit is used for generating executable data query sentences based on the standard business terms, the query intention and a preset query template; and the execution output unit is used for executing the data query statement to obtain a data result, converting the data result into a natural language reply and outputting the natural language reply.
  9. 9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the intelligent question-answering method according to any one of claims 1 to 7 when the computer program is executed by the processor.
  10. 10. A computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements the steps of the intelligent question-answering method according to any one of claims 1 to 7.

Description

Intelligent question-answering method, device, equipment and medium Technical Field The invention relates to the technical field of artificial intelligence, which can be applied to the fields of financial science and technology and medical health, in particular to an intelligent question-answering method, device, equipment and medium. Background Along with the deep digital transformation of enterprises, the requirement of realizing intelligent question and answer by utilizing an artificial intelligence technology is particularly urgent in the fields of financial science and technology, medical health and the like. Currently, many enterprises attempt to deploy question-answering systems based on large language models in an effort to allow employees to directly obtain data results through natural language questions. However, this technology presents significant challenges in a practical enterprise environment. Firstly, answer consistency cannot be guaranteed, and queries for the same business term (such as "sales") may get different calculation results due to the context difference due to uncertainty of large model generation, resulting in confusion of decision basis. Secondly, there is a gap between business semantics and technical implementation, and ambiguous expressions (such as "recent performance") in natural language are difficult to be accurately resolved into specific indexes, dimensions and time ranges, so that query failure or result deviation is caused, and the accuracy is low. Disclosure of Invention The invention provides an intelligent question-answering method, an intelligent question-answering device, computer equipment and a medium, which are used for solving the technical problem that the accuracy of the existing intelligent question-answering is low. In a first aspect, an intelligent question-answering method is provided, including: Receiving a natural language question input by a user, and processing the natural language question to obtain a service entity and a query intention; matching and disambiguating the service entity with term nodes in a pre-constructed service term library to determine standard service terms corresponding to the service entity in the service term library; generating executable data query sentences based on the standard business terms, the query intents and preset query templates; executing the data query statement to obtain a data result, converting the data result into a natural language reply and outputting the natural language reply. In a second aspect, an intelligent question answering device is provided, including: the receiving and processing unit is used for receiving natural language questions input by a user and processing the natural language questions to obtain service entities and query intents; The matching disambiguation unit is used for matching and disambiguating the service entity with term nodes in a pre-constructed service term library so as to determine standard service terms corresponding to the service entity in the service term library; the generation unit is used for generating executable data query sentences based on the standard business terms, the query intention and a preset query template; and the execution output unit is used for executing the data query statement to obtain a data result, converting the data result into a natural language reply and outputting the natural language reply. In a third aspect, a computer device is provided, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the intelligent question-answering method described above when the computer program is executed by the processor. In a fourth aspect, a computer readable storage medium is provided, the computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of the intelligent question-answering method described above. In the scheme realized by the intelligent question-answering method, the intelligent question-answering device, the computer equipment and the storage medium, natural language questions input by a user can be received, the natural language questions are processed to obtain service entities and query intention, the service entities are matched and disambiguated with term nodes in a pre-built service term library to determine standard service terms corresponding to the service entities in the service term library, executable data query sentences are generated based on the standard service terms, the query intention and a preset query template, the data query sentences are executed to obtain data results, and the data results are converted into natural language answers and output. The invention firstly processes the natural language problem to obtain a service entity and a query intention, then matches and disambiguates the service entity with term nodes in a pre-constructed service term library to