Application of BERT-based local optimization algorithm in intelligent question answering system design

Abstract

With the rapid growth of artificial intelligence, especially deep learning, Natural Language Processing (NLP) is playing an increasingly important role in human learning, work and life. The huge amount of information makes it difficult for people to get valuable information from it. How to quickly screen this valuable information is the key to the intelligent question answering system. In order to obtain knowledge information that is more in line with users' expectations, NLP technology has been researched and developed, and related technical products have been successfully integrated into people's lives, among which intelligent question answering system can better meet people's demand for accurate information. This article explains some limitations and defects of current semantic matching technology, puts forward a local optimization algorithm based on Bert, and applies it to the design of intelligent question answering system. The simulation results show that this algorithm is more accurate for text feature recognition, which is 19.85% higher than the contrast algorithm. The system interacts with users through the visual interface, and automatically replies to the questions raised by users, thus achieving the purpose of practical application.