Ontology-based Hazard Information Extraction From Kazakh Food Complaint Documents

Food is a basic material for human being to survive and develop. Food safety is a closelyrelated to people’s health and safety. In order to solve the food safety problem, it is necessaryto introduce an independent system of supervision by public opinion. Consumers come tochoose internet, which has features of low cost and fast spread speed. In fact internet wasleading power of supervision by public opinion in many food safety accidents in recent years.Information extraction from Kazakh complaint document refers to extract relatedharmful information from plentiful Kazakh complaint document, as same thematic complaintdocument are kept dispersedly in different website and its forms have nothing in commonwith each other. We propose a model of ontology-based hazard information extraction fromKazakh food compliant documents based on the typical model of ontology-based informationextraction. It not only can accurately extract hazard information from Kazakh food complaintdocuments but also can reason about the semantic meaning of the hazard information. Thenthe hazard information is used to perfect the ontology in order to keep the ontology real-time.The intelligent information processing technology is an important safeguard on the netsupervision of food safety.There are two main components in our information extraction model. One is theLearning Model; the other is the Extraction Model. In the learning model we will completethe tasks of seed word generation and related words generation. That will pave the way of thefollow-up tasks of extracting hazard information from Kazakh food complaint documents. Inthe extraction model, we will extract the hazard information from food complaint documents.There are three types of information “background knowledge”,“negative words” and “hazardinformation”. The integration of these three types of information can not only solve theproblem of information fragments effectively, but also let consumer clearly see the hazard offood. In the process of extracting hazard information from Kazakh food complaint documents,we take advantage of meaning background analyzed by ontology to choose diary productscomplaint document，then use seed words to do second extraction of the complaintdocument, examining harmful information in the complaints document. If it does have suchinformation, output the information into semantic date base.