Automatic Recognition Of Semantic Roles In Mongolian Adjective Phrases

As a branch of modern linguistics,quantitative linguistics seeks to explore the structural patterns and evolutionary laws of language through quantitative methods.The three characteristics of quantitative linguistics are "precise,real and dynamic",so this paper constructs a corpus based on real language materials and provides a reliable language for quantitative analysis of the semantic roles of adjectives and phrases.The semantic role tagging research is one of the shallow semantic analysis contents at the sentence level.With the continuous progress and development of Mongolian information processing research,it is in the critical period of sentence processing and semantic processing.Based on the self-built multi-layer tagged corpus,this paper uses the theory and methods of quantitative linguistics to quantitatively analyze the semantic role characteristics of adjective phrases,and conducts automatic recognition research on the semantic role of adjective phrases.This paper consists of five parts: introduction,construction of Mongolian adjective phrase information database,adjective phrase semantic role feature measurement study,adjective phrase semantic role recognition experiment and result analysis,conclusion and outlook.The first chapter,mainly introduces the significance of the study,the background of the study,the content of the study,and the research methods and steps.The second chapter,the steps of constructing the Mongolian Adjective Phrase Information Database are introduced.Based on the 10,000-sentence scale multi-layer tagged corpus,the Mongolian Adjective Phrase Information Database,which includes9 attribute field values,is constructed.2,646 adjective phrases are included in the database,of which 1,047 adjective phrases play 17 semantic roles in the multi-layer tagged corpus.The third chapter,a semantic role feature measurement study of adjective phrases is conducted.Using the values of the attribute fields of the Mongolian Adjective Phrase database,such as number of words,internal structure relationship of phrases,semantic classification,sentence constituents,and semantic roles,the measurement analysis is conducted in terms of the distribution characteristics of the semantic roles of adjective phrases,length distribution characteristics,distribution characteristics of internal structure relationship of phrases,correspondence with sentence constituents,semantic classes of semantic roles of adjective phrases,semantic role frame changes of adjective phrases,and dominant words of adjective phrases In the fourth chapter,the analysis is carried out.The fourth chapter,the steps of the adjective phrase recognition experiment and the analysis of the results are explored.On the basis of the measurement study of the semantic role features of adjective phrases,675 rules for semantic role recognition of adjective phrases were summarized using a rule-based complex feature description method,and an automatic recognition program for semantic roles of adjective phrases was written.The research results show that the the correct rate of theme adjective phrase semantic role recognition is 74.35%,and the recognition rate of the remaining16 adjective phrase semantic roles is low.The fifth chapter,summarizes the research results of this paper and the research content that needs to be added in the later stage.The research results achieved in this paper are beneficial to further understanding the semantic features of adjective phrases,and will fill the gap in the research of automatic semantic role recognition of adjective phrases.