Multi-algorithm model fused FAQ question and answer matching method and system

Abstract

The invention provides an FAQ question and answer matching method and system fused with a multi-algorithm model, and the method comprises the steps: constructing an inverted index based on an FAQ knowledge base; obtaining a user question, performing preprocessing to obtain a first preprocessing result, and searching a knowledge base question associated with the user question in the reverse index; training a sentence encoder, encoding the user question and the knowledge base question into sentence vectors by using the sentence encoder, and calculating a cosine value, namely similarity, between the sentence vectors of the user question and the knowledge base question; preprocessing the user question and the knowledge base question to obtain a second preprocessing result; calculating the matching degree of the second preprocessing results of the user question and the knowledge base question by using a matching algorithm; performing weighted average calculation by combining the similarity and the matching degree to obtain a final similarity value; and giving an answer according to the final similarity value and a set threshold value. According to the method, algorithm logic models based on different rules are embedded, so that the overall optimal matching effect is achieved, the overall question and answer accuracy is high, and the effect is good.

Description

technical field [0001] The present invention relates to the technical field of natural language processing, and in particular, to a method and system for FAQ question-and-answer matching that integrates a multi-algorithm model. Background technique [0002] In the intelligent customer service system, the question and answer based on FAQ (frequently-asked questions, referred to as FAQ) is the basic ability of the system, and the accuracy of the FAQ question and answer is a very important indicator to measure the advanced nature of the system. Based on the characteristics of different industries and different business priorities of customers, the maintenance level of the FAQ database is different. [0003] At present, FAQ matching is generally carried out by using models or rules, and this matching method has great limitations. (1) Limitations of the model: The model itself needs to be trained well before it can be used. There will be a lag in online business changes. If ther...

AI Technical Summary

Problems solved by technology

(1) Limitations of the model: the model itself needs to be trained well before it can be used, and there will be a lag in online business changes. If there is a badcase online, it will be slow to repair

(2) Limitations of the rules: The rule algorithm will rely too much on basic semantic data, such as similar questions, words and other data. Once these words have problems with maintenance, it will lead to algorithm errors

Images