Research On Phrase Recognition Based On Probability Analysis And Rule Constraints

Nowadays,on the background of big data,massive data requires the support of natual language processing technology in order to find the valuable information effectively.For this reason,the development of natural language processing(NLP)is an inevitable trend.The research of phrase recognition is an important and basic task in the field of NLP,which belongs to shallow analysis.The divide-and-rule method of shallow analysis helps a lot for the disambiguation of complete syntactic parsing.Thus,it is full of value and significance to study phrase recognition.This paper attempts to discuss a more general model to obtain and recognize phrase for the general natural language in theory.The main work includes:(1)Based on the method of probability analysis and rule constraints,we put forward the concept of Combination Degree to expound how to obtain and recognize phrase.(2)In chapter 5,we prove out our method by obtaining and recognizing phrasal verbs in English language which focus on both separable and inseparable phrasal verbs.It is by adopting several ways,such as corpus training,Conbination Degree,word similarity,data smoothing,rule constraints and mimetic phrasal verb lexicon,to extract English phrasal verbs in detail.The programming language of our system is Java,and the test set is analyzed by the platform of Java Web.(3)According to the experimental results,the precision can achieve 88%and the recall precision can achieve 90%,which shows that this method is effective and feasible.In conclusion,the innovation of the paper is as follows:(1)Based on the method of probability analysis and rule constraints,this paper presents the concept of Combination Degree to explore how to obtain and recognize phrase in general natural language.(2)The word similarity is applied to data smoothing.(3)Dynamic corpus can be achieved in the system.