Search

CN-117435702-A - Educational field question-answering method, educational question-answering integrated machine and storage medium

CN117435702ACN 117435702 ACN117435702 ACN 117435702ACN-117435702-A

Abstract

The invention relates to the technical field of artificial intelligence and provides an educational field question-answering method, an educational question-answering integrated machine and a storage medium, wherein the method comprises the steps of firstly obtaining the current problem of a user aiming at the educational field; then based on the educational question-answering model, the intention of the current question is identified by adopting the current interaction history, the answer strategy of the intention is determined based on the category of the intention, and the current question is answered based on the intention and the answer strategy thereof. The reply strategy comprises calling a plug-in reply strategy, inquiring a card reply strategy or generating a reply strategy. The method can accurately analyze the questions presented by the user by identifying the intention of the current questions, and is beneficial to accurately replying the current questions. In addition, the method determines the answer strategy of the intention by utilizing the category of the intention, and then replies to the current question, so that different answer strategies can be provided for questions with different intentions in a targeted manner, the accuracy of the answer is ensured, and the experience of a user is improved.

Inventors

  • WANG SHIJIN
  • SONG ZIYAO
  • GONG JIEFU
  • SHENG ZHICHAO
  • WEI SI
  • HU GUOPING

Assignees

  • IFLYTEK CO LTD
  • IFLYTEK BEIJING CO LTD

Dates

Publication Date
20240123
Application Date
20231023
Priority Date
20231023

Claims (10)

  1. 1. An educational field question-answering method, characterized by comprising: acquiring the current problem of the user aiming at the education field; identifying the intention of the current question by adopting a current interaction history based on an educational question-answering model, determining an answer strategy of the intention based on the category of the intention, and replying to the current question based on the intention and the answer strategy thereof; the reply strategy comprises calling a plug-in reply strategy, inquiring a card reply strategy or generating a reply strategy.
  2. 2. The educational field question-answering method according to claim 1, wherein answering the current question based on the intention and its answer policy, comprises: If the current interaction history is not empty and the intention is incomplete, rewriting the intention by adopting the current interaction history to obtain a complete intention; and replying to the current question based on the complete intention and the corresponding reply strategy thereof.
  3. 3. The educational field question-answering method according to claim 2, wherein the intent includes a plurality of; determining a reply strategy of the intention based on the category of the intention, and replying to the current question based on the intention and the reply strategy thereof, wherein the method specifically comprises the following steps: if the current interaction history is not null, a first intention and a second intention exist in a plurality of intentions, a dependency relationship exists between the first intention and the second intention, and the first intention and the second intention are incomplete, determining a reply strategy of the first intention based on the first intention and a reply strategy of the second intention based on the second intention respectively based on the education question-answering model, rewriting the first intention by adopting the current interaction history to obtain a first complete intention, replying the first complete intention based on the reply strategy of the first intention, rewriting the second intention based on the reply result of the first complete intention to obtain a second complete intention, and replying the second complete intention based on the reply strategy of the second intention.
  4. 4. The educational field question-answering method according to claim 2, wherein the current question carries a business identification; and adopting the current interaction history to rewrite the intention to obtain a complete intention, wherein the method comprises the following steps: determining intention rewriting prompt information corresponding to the service identifier based on an interaction configuration loading module of the education question-answering model; and determining the complete intention by adopting the intention rewriting prompt information and the current interaction history based on the intention rewriting module of the education question-answering model.
  5. 5. The educational field question-answering method according to claim 2, wherein the current question carries a business identification; replying to the current question based on the complete intent and its corresponding reply policy, including: based on the interaction configuration loading module of the education question-answering model, determining instruction parameters corresponding to the service identification to generate prompt information; the instruction parameter generation module based on the education question-answering model adopts the complete intention and the answer strategy corresponding to the complete intention, and the instruction parameter generation prompt information to generate the instruction parameter of the answer strategy corresponding to the complete intention; And executing a reply strategy corresponding to the complete intention by adopting instruction parameters based on an execution module of the educational question-and-answer model to obtain the answer of the current question.
  6. 6. The educational field question-answering method according to claim 5, wherein based on the execution module of the educational question-answering model, executing the answer strategy corresponding to the complete intention with instruction parameters, obtaining the answer of the current question, previously comprising: the instruction parameters are verified based on a verification module of the education question-answering model; correspondingly, based on the execution module of the educational question-answer model, executing the answer strategy corresponding to the complete intention by adopting instruction parameters to obtain the answer of the current question, wherein the answer comprises the following steps: if the verification is passed, executing a reply strategy corresponding to the complete intention by adopting the instruction parameters based on the execution module to obtain an answer of the current question; if the verification fails, a dialogue is established with the user based on the parameter attribute of the reply strategy corresponding to the complete intention, the parameter information of the reply strategy corresponding to the complete intention is determined based on the reply content of the user, and the reply strategy corresponding to the complete intention is executed by adopting the parameter information based on the execution module, so that the answer of the current question is obtained.
  7. 7. The educational ait question-answering method according to any one of claims 1-6, wherein the current question carries a business identification; identifying an intent of the current question using the current interaction history, including: determining intention recognition prompt information corresponding to the service identifier based on an interaction configuration loading module of the education question-answering model; and based on the intention recognition module of the education question-answering model, recognizing the intention of the current problem by adopting the intention recognition prompt information and the current interaction history.
  8. 8. The educational ait question-answering method according to any one of claims 1-6, wherein the current question carries a business identification; determining a reply strategy of the intention based on the category of the intention, comprising: based on the interaction configuration loading module of the education question-answering model, determining function planning prompt information corresponding to the service identifier; and a function planning module based on the education question-answering model, which adopts the function planning prompt information and the current interaction history to identify the category of the intention, and determines the answer strategy of the intention based on the category of the intention.
  9. 9. An educational question-answering machine comprising a memory, a processor, an educational question-answering model stored on the memory, and a computer program stored on the memory and executable on the processor; the processor, when executing the computer program, invokes the educational question-answering model to implement the educational domain question-answering method according to any one of claims 1-7.
  10. 10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the educational field question-answering method according to any one of claims 1-7.

Description

Educational field question-answering method, educational question-answering integrated machine and storage medium Technical Field The invention relates to the technical field of artificial intelligence, in particular to a question-answering method in the education field, an educational question-answering integrated machine and a storage medium. Background With the development of deep learning technology, the field of natural language processing is also gradually rising, wherein a dialogue system has wide application scenes in life and good prospect, and is a popular natural language processing task at present. Current dialog systems are involved in a number of areas, including e-commerce platform customer service, reservation of movie tickets, medical consultation, and the like. In the education field, in order to enable students to have better knowledge acquisition channels, online knowledge questions and answers are also developed gradually. The on-line knowledge question-answering system can adopt a deep learning model to carry out semantic understanding and knowledge retrieval according to natural language questions presented by students and return answers to the students in a natural language form. However, due to the diversity of student question expressions, the resolving power of the existing on-line knowledge question answering system for complex questions is still poor, and answers obtained after the questions are resolved are not accurate enough, so that the answer is wrong, and the effect is poor. Disclosure of Invention The invention provides a question-answering method in the education field, an educational question-answering integrated machine and a storage medium, which are used for solving the defects existing in the prior art. The invention provides a question-answering method in the education field, which comprises the following steps: Acquiring the current problem of the user aiming at the education field; identifying the intention of the current question by adopting a current interaction history based on an educational question-answering model, determining an answer strategy of the intention based on the category of the intention, and replying to the current question based on the intention and the answer strategy thereof; the reply strategy comprises calling a plug-in reply strategy, inquiring a card reply strategy or generating a reply strategy. According to the method for answering questions in the education field provided by the invention, the current questions are answered based on the intention and the answer strategy thereof, and the method comprises the following steps: if the current interaction history is not empty and the intention is incomplete, rewriting the intention by adopting the current interaction history to obtain a complete intention; and replying to the current question based on the complete intention and the corresponding reply strategy thereof. According to the educational field question-answering method provided by the invention, the intention comprises a plurality of questions; determining a reply strategy of the intention based on the category of the intention, and replying to the current question based on the intention and the reply strategy thereof, wherein the method specifically comprises the following steps: if the current interaction history is not null, a first intention and a second intention exist in a plurality of intentions, a dependency relationship exists between the first intention and the second intention, and the first intention and the second intention are incomplete, determining a reply strategy of the first intention based on the first intention and a reply strategy of the second intention based on the second intention respectively based on the education question-answering model, rewriting the first intention by adopting the current interaction history to obtain a first complete intention, replying the first complete intention based on the reply strategy of the first intention, rewriting the second intention based on the reply result of the first complete intention to obtain a second complete intention, and replying the second complete intention based on the reply strategy of the second intention. According to the education field question-answering method provided by the invention, the current question carries a service identifier; and adopting the current interaction history to rewrite the intention to obtain a complete intention, wherein the method comprises the following steps: determining intention rewriting prompt information corresponding to the service identifier based on an interaction configuration loading module of the education question-answering model; and determining the complete intention by adopting the intention rewriting prompt information and the current interaction history based on the intention rewriting module of the education question-answering model. According to the education field question-answering method provided by