Ontology-based Semantic Similarity Calculation Nutrition Quiz System

The information on the network increases and updates faster and faster. At the same time there is a lot of repetitive spam, which causes some inconvenience for user to get effective information. On one hand, traditional search engine only returns a large number of relevant web pages or documents, on the other hand, the information retrieval techniques based on keywords have difficulty in meeting users’ information needs in the semantic level and knowledge level. The ontology have good hierarchy structure of concepts and it is powerful to express concepts’semantics, so ontological technology has become an important modeling tool of knowledge in the field of information science and knowledge management. Question answering system based on semantic comprehension can effectively compensate these defects of traditional search engines, which has been proven. This thesis uses ontological technology, and domain ontologies are used as knowledge resource for semantic understanding and processing of users’question.This thesis analyzes and researches the taxonomic relationship between concepts in the ontology and several traditional ontology-based concept similarity calculation models, and it also discusses several impact factors in the similarity calculation models based on ontology. This thesis proposes and implements an improved semantic similarity model based on concept’s information content. Compared with the conventional calculation models of concepts’information content, the improved model takes the number of concepts’ hypernyms and the number of concepts’ leaf nodes as the main calculation factors and takes concepts’ relative depth as a weight coefficient, so the new model can distinguish different concepts by identifying concepts’ topology. Based on our proposed model, a new based-ontology semantic similarity calculation model is implemented. Experiments show that the similarity model implemented by this thesis can capture concepts’ semantic information hidden in the classification hierarchy of concepts effectively. A food ontology and an FAQ of nutrition knowledge are built as the knowledge resources in the question answering system. Combined with word segmentation and semantic extension, the thesis finally designs and implements a question answering system based on FAQ, which can provide the nutrition and food knowledge in a fast and easy way.