Event Trigger Recognition Based On Positive And Negative Weighting And Its Application

Event,as the basic unit of human knowledge,has always been favored by researchers in many fields.In recent years,with the rapid development of Internet,new media has gradually become the main way for people to obtain information.At the same time,there are frequent emergencies,and the cyberspace is full of massive electronic text data containing various types of emergency information.The effective use of event information on electronic text can provide important support for information retrieval,public opinion analysis,text topic classification and other research fields,making event extraction technology become a research hotspot.Event trigger recognition is a sub-task of event extraction.At present,there are problems of insufficient effective auxiliary features and insufficient utilization of auxiliary features when using word features as a benchmark,resulting in the final event trigger recognition effect is not ideal.In order to solve these two problems,this paper proposes an event trigger recognition method based on positive and negative weighting,and on this basis,the application research of text topic classification based on event trigger is carried out.The main research work includes the following two parts:(1)Research on event trigger recognition based on positive and negative weighting.In order to improve the effect of event trigger recognition when using word features as a benchmark,a positive and negative weighting event trigger recognition method is proposed by constructing a trigger table with positive and negative features and combining with positive and negative weighting algorithm.First of all,a feature called associated word is defined by combining the left and right position words and the parent node in dependency parsing,which is helpful to improve the effect of benchmark method.Then,the single feature is divided into positive or negative feature,and combined with positive and negative weighting algorithm to assist the benchmark method for event trigger recognition,improving the role of the single feature in event trigger recognition.Finally,we combine multiple features together.The combination of these features with the positive and negative weighting algorithm is used to assist the benchmark method for event trigger recognition,which further improves the effect of event trigger recognition method in this paper.The experimental results show that the event trigger recognition method based on positive and negative weighting has achieved satisfactory effects.(2)The application research of text topic classification based on event triggers.Firstly,based on the idea of Term Frequency-Inverse Document Frequency,a method called Trigger Frequency in Category-Inverse Category Frequency is proposed to calculate the feature weight between trigger feature and text topic category.Then,aiming at the difference in correlation between event triggers and text topic categories,the chi-square test method is improved to reflect the positive and negative correlation.And it is used as the weight factor to multiply the feature weight to complete the calculation of feature weight.Finally,the feature weight between each feature and the text topic category in the text is accumulated according to the text topic category.And the topic category with the largest accumulated value is taken as the text classification result,so as to complete the text topic classification.Experimental results show that the text topic classification method based on event triggers has achieved ideal effects.Figure[15]Table[12]Reference[65]...