Construction Of Mongolian Predicate Information Database

In this paper, based on the phrase structure grammar, case grammar and other theories of modern linguistics, semantic structure information of the Mongolian sentence are labeled. With these approaches, the Mongolian predicate information database was constructed and the relevant preliminary statistics was collected. First of all, statistical analysis of the sentence length have been adopted for the one million words-level modern Mongolian corpus. Secondly, for the specific needs of Mongolian information processing, the5,000simple sentence length between3to20words was selected, and6kinds of information, such as fixed phrases, part of speech, phrase structure, semantic classification, syntactic component, semantic roles were labeled and "Mongolian predicate information database" was constructed. Finally, by using the predicate information database, statistical analysis of the relevant information, category, morphology, phrase structure, words number and semantic features of Mongolian predicate was introduced, and preliminary statistical research on Mongolian predicate information was developed. The structure and main contents are as follows:The introduction part, mainly introduces the meaning, research survey, theories and methods, data sources and marking instructions.The first chapter introduces the selecting sentences, statistic rules, selection criteria and the sentence tagging rules.The second chapter introduces the process of database construction and property fields list.The third chapter we analyze the data collected from real corpus including the category, morphology, phrase structure, words number and semantic features and so on.The conclusion part summarizes the whole thesis and puts forward the next research task.