Research On Ambiguity Processing Based On Statistics And Knowledge And Its Application In The Design Of Intelligent Instruments

Natural language understanding is an important aspect of artificial intelligence. And ambiguity processing is one of the difficult points in natural language understanding. Based on the analysis of the phenomenon of ambiguity, this paper puts forward the method of eliminating ambiguity based on the statistics and knowledge of the special structure of the collocation. And combined with dependency syntactic parsing, this paper establishes the disambiguation model and applies the ambiguity processing model of the special structure of the collocation to the design of intelligent instrument. Finally realize the intelligent instruments by understanding the requirements of users.This paper is aimed at some common fixed collocation structure, because they are not adjacent to the position, resulting in the segmentation ambiguity of the distance. For the modern Chinese, because some sentences have clauses, it is easy to be ambiguous when the sentence is to determine the subject. So this paper makes a classification of the special collocation structure ambiguity in the use of statistical probability processing, which is divided into the crossing ambiguity of the preposition and noun of locality in the distance, the combination ambiguity of the preposition and the verb in the distance, the crossing ambiguity of the verb and the preposition in the distance and the collocation structure ambiguity of the subordinate clause. Firstly, for these types, we use the method of statistical disambiguation to deal with the ambiguous sentences and determine the initial segmentation results by using mutual information rate. Then we analyze the structure of the sentence through dependency syntactic parsing to judge whether the sentence is in accordance with the syntax. For the segmentation ambiguity of the distance, when there is ambiguity, it is easy to make a mistake in the sentence structure. Syntactic parsing can correctly judge the sentence structure, so as to rule out the wrong way and eliminate ambiguity. Finally, we deal with the results after the syntactic parsing based on the disambiguation of knowledge and analyze the subject concept of sentence events whether it matches the knowledge base template or not. The collocation structure ambiguity of the subordinate clause due to the existence of clause, when the event in the sentence in the determination of the main body, prone to ambiguity, so it is more need to eliminate ambiguity based on knowledge.In this paper, we establish the disambiguation model based on statistics and knowledge. The feasibility of the model is verified by combining with the above four kinds of special structure of the collocation ambiguous examples. According to the overall ambiguity elimination model combined with the specific examples in the field of intelligent instruments, the disambiguating model based on the combination of statistical and knowledge is applied to the design of intelligent instruments in order to deal with the above four types of ambiguity phenomenon in the intelligent instrument system. So it can improve the requirement analysis ability of instruments.