Studies On Automatic Recognition Of Chinese Preposition’s Usages And Application On Information Extraction

As a type of function words, prepositions does not have specific lexical meaning like content words, but prepositions and prepositional phrases bear the heavy syntactic task, and play a vital role in Chinese syntax analysis and semantic understanding. One preposition may indicate a different meaning and have different usage in different context. Therefore, in the paper, we have done a detailed research on preposition’s usages, and searched the law of preposition’s usages with artificial summarize and machine learning methods. Then we devised the formal description to preposition’s usages and build the preposition’s usages knowledge base. On the basis of preposition’s usages knowledge base, the paper conducted three studies which are the automatic recognition on Chinese preposition’s usages based on rules, the automatic recognition on Chinese preposition’s usages based on statistics and the automatic recognition on Chinese preposition’s usages based on rules and statistics. In the experimental results, the precision rate of the automatic recognition on Chinese preposition’s usages based on rules is67.38%, the precision rate based on statistics is76.80%, there are two combination methods of rules and statistics, the precision rate of macro combination is78.47%,11.09%higher than the rules,1.67%higher than the statistics, the precision rate of micro combination is82.02%,14.64%higher than the rules,5.22%higher than the statistics. Therefore, the micro combination method of rules and statistics is the best way to automatic recognition in this paper.It is possible for prepositions to be applied in related areas through the study of preposition’s usages and preposition’s usages automatic recognition. This paper proposes an event information extraction method based on the preposition’s usages. Since preposition can mark the sentence structure and constituent commendably, it clarifies the sentence’s information and is helpful for event information extraction. In this paper’s experiment results, when experiment corpus is original, the precision rate, recall rate and F values were71.71%,69.65%and70.66%respectively, when experiment corpus is recognized by human, the precision rate, recall rate and F values were92.24%,89.83%and91.02%respectively, when experiment corpus is recognized by machine, the precision rate, recall rate and F values were90.88%,86.92%and88.86%respectively. Form the results, the research of preposition’s usages is valuable in information extraction.Finally, the paper summarized the research and proposed the next step, pointed out that building preposition knowledge base is a milestone in research on preposition. At the same time, the study of preposition also provides services for information extraction technology, and the paper clarifies the next step of the research.