Research On Food Complaint Document Classification Based On Semantic Kernel Function

The food is a necessity of human existence,is the material basis for humansurvival and development.Food safety is a major livelihood issues,and in recentyears,has been producing and selling fake or substandard public exposure,endlesscases of toxic and hazardous food,tens of thousands of victims,food safety problemsoccur more frequently caused widespread concern in the community.With the rapiddevelopment of Internet and the improved food security monitoring system,more andmore people through online complaints to participate in food safety management.Butthe food problem people posted online complaints text is uneven,there is no effectiveorganization,and different users have different needs for complaint information.Howto organize and manage these massive complaints text which can quickly,accuratelyfind the information they need,to solve the current problems in the field of computerscience.Text classification technology is an important foundation of information retrievaland text mining,is often used in text organization and processing of large,so that theinformation be orderly,convenient for people to find.However,many chinesetraditional text classification algorithms are numerical information through thestatistical characteristics of text classification,only consider the characteristics of thefrequency,and ignore the text feature semantic correlation between text and chinesesyntactic structure.According to the characteristic chinese text,this paper presents achinese semantic kernel food based on domain ontology,to join the "Hownet"similarity computation based on string kernel function,solves the synonyms and nearsynonyms chinese text in question,but without the knowledge base to limit the casecan also chinese text semantic information the embodiment of good;converting thecalculation of text similarity is the similarity between the tree structure representationof text using tree kernel function,can very well reflect the chinese grammar structureinformation.Both are combined to form a relatively perfect chinese semantic kernelfunction, in order to achieve a better Chinese text classification.