Similarity Computing Method of Poly-Encoders Short Texts in Both Chinese and Lao Based on Part of Speech Characteristics

Similarity computing of short texts in both Chinese and Lao is a basic task in bilingual research of the Chinese and Lao, which is widely used in the fields of text matching and QA question answering. The corpus resources of the Lao language are scarce, and the research results by current methods are poor. Therefore, we propose a similarity calculation method by incorporating bilingual language features and combining the common part-of-speech features of Chinese and Lao bilinguals to obtain more semantic information from the limited training corpus. At the same time, we get the text-matching information by the Poly network to further improve the accuracy of semantic vector representation, and finally, the similarity calculation model of Chinese and Lao bilingual short texts is constructed. Experimental results show that the proposed method achieves better results than the existing methods (74.95%).