Title: Research on semantic similarity calculation methods in Chinese financial intelligent customer service

Authors: Mingyu Ji; Xinhai Zhang

Addresses: College of Information and Computer Engineering, Northeast Forestry University, Harbin 150040, Heilongjiang, China ' College of Information and Computer Engineering, Northeast Forestry University, Harbin 150040, Heilongjiang, China

Abstract: Semantic similarity computing is an important task in NLP, which aims to calculate the semantic similarity between texts. In order to solve the problem of low accuracy of existing calculation methods in Chinese financial intelligent customer services, we propose a text similarity calculation method combining Capsule-BiLSTM network and Chinese Part of Speech (POS) correction. In the text pre-processing stage, POS correction is performed on financial professional words and ambiguous words in the data set to reduce the influence of Chinese word segmentation errors on similarity judgments. Capsule network and BiLSTM were used to obtain local and global information, respectively, the two similarity matrices obtained are fused to determine the similarity. Experimental results show that the proposed method compared with other conventional methods has high precision and F1-value in ATEC data set.

Keywords: semantic similarity; capsule network; BiLSTM; part of speech correction.

DOI: 10.1504/IJCAT.2022.123465

International Journal of Computer Applications in Technology, 2022 Vol.68 No.2, pp.143 - 149

Received: 18 Apr 2021
Accepted: 03 Jun 2021

Published online: 22 Jun 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article