NLP-empowered Extraction And Application Of Semantic Frames And Constructions

The COVID-19 pandemic is a global event that deeply influences every aspects of human society and results in fundamental change of people’s life for two years.Language as an essential media for us to exchange concepts is also influenced by the COVID-19 situation and adaptively reflects how human beings encode newly emerged situation and concepts to it.News publishers offers large amount of pandemic related text data about that are open accessible on the internet.News data are valuable resources for linguistic researches for the language information it encodes in a relatively well written manner.This research adopts a cognitive linguistic point of view for analyzing language phenomena aroused by COVID-19,specifically,the frame semantics’ and construction grammar’s language philosophy.These two theories are inter-connected and complementing each other in terms of their approaches of interpreting language data,which jointly offers a comprehensive description of language use in the domain.Linguistic analysis structured by these two theories can be beneficial for language teaching and learning.The biggest challenge this study face is to identify and extract semantic frames and their related grammatical constructions in a corpus consisting more than 80 million words that are unstructured.The solution this study that are unstructured.The solution this study proposes is to utilize state-of-the-art natural language processing(NLP)techniques and to engineer them into a computational pipeline designed on the basis of theoretical analysis of the characteristic of semantic frame and grammatical constructions.The core of the solution is to create meaning clusters in word and sentence level with embedding techniques for the purpose of distilling out and categorizing the meaningful forms which can potentially be taken as constructions.