Question and answer retrieval method based on campus business

Abstract

The invention discloses a question and answer retrieval method based on campus business. The method comprises the following specific steps: 1, a student enters a user interface; 2, the student inputs and triggers an input mechanism; 3, a retrieval process is executed, and the system calculates scores of all documents by using a BM25 algorithm according to key sentences / keywords in student query; and 4, the documents with the highest score are reported to the student as an answer. An offline process is introduced, a feedback mechanism is formed, updating of a knowledge base is accelerated, and better user experience is provided for students. When the students input, four input mechanisms are introduced, input is changed into clicking, and the load rate and the error rate are reduced. On the basis of campus background knowledge, a BM25 retrieval algorithm is improved, weights are given to documents corresponding to departments, and answers can be screened out more efficiently and accurately.

Description

Technical field: [0001] The invention relates to the technical field of information retrieval (Information Retrieval) methods, in particular to a question and answer retrieval method based on campus business. Background technique: [0002] With the rapid development of deep learning technology, artificial intelligence has been widely used in life and work, and is committed to improving people's lives. The natural language question answering system is one of the products of artificial intelligence. The question answering system is popular in various fields, such as finance, justice, etc. There are also question answering systems on campus. Students often use the question answering system to quickly solve problems. [0003] The question answering system includes a variety of natural language processing technologies, among which question answer retrieval is a crucial technology. [0004] At present, many question-and-answer retrieval algorithms or models used in campuses have ...

AI Technical Summary

Problems solved by technology

[0004] At present, many question-and-answer retrieval algorithms or models applied to campuses have greatly improved the accuracy compared with traditional algorithms, but these algorithms or models only focus on the information that the question and answer text itself can provide, ignoring the information based on The background knowledge of the campus, but only part of the background knowledge can be obtained from the text information of the questions and answers, resulting in the algorithm not fully "understanding" the background knowledge of the campus, which makes the matching between the question and each answer poor. low accuracy

In this way, students often cannot get the answers they need through the question answering system, and the actual experience is poor.

Images