A Concise Chinese Text Time Relationship Recognition Scheme
Wenfeng Hu, Chen Xiangwei · 2022
Temporal relationship recognition plays an important role in many areas of natural language processing. The current mainstream research aspect in the related field is guided by the approach of English relationship recognition, focusing on event-time-relationship recognition. In view of the high complexity presented by the event-time-relationship recognition task and the differences in the grammar of Chinese and English texts, this paper no longer adopts the idea of extracting events first and matching time later, and proposes a processing scheme for temporal slicing of Chinese documents. The paper first summarizes the extensive research on Chinese time marker recognition and normalization, and designs a more reliable set of rule-based solutions; then the documents are time-sliced through the identified time markers to obtain document time slices. The algorithm is tested on the data and achieves better results on news texts.