State Evaluation And Evolution Analysis Of Tourism Emergency Based On Domain Knowledge

As the globalization degree and the economic strength are enhanced, the tourism industry develops rapidly which has become an indispensable and significant industry in our national economy. However, in the context of globalization, more and more uncertain factors lead to the occurrence of tourism emergencies and a great threaten to life and property safety of tourists. Compared with other industries, the tourism industry is more sensitive to emergencies, which means that emergency events have huge destructiveness and influence on it. In recent years, a large amount of information materials related to tourism emergencies have been accumulated, including the past cases, preplans, laws and policies, expert knowledge and experience, and so on. Thus, according to the existing information, we can build the domain knowledge of tourism emergency, and study its occurrence, development and evolution on such basis. It can not only improve the ability of emergency management, but also effectively reduce the loss of life and property for tourists.The main research work and innovations are as follows:(1) On the basis of Internet data information of tourism emergency, we collect the domain terms and concepts, define the concepts and their relationship of the domain ontology (mainly including the hierarchy, classification, evolution relation, etc.), give the attributes and the constraint conditions of concepts, establish the ontology model of the domain knowledge from the aspects of event attributes, event life cycle, event evolution relation, eliminate the ambiguity of terms and concepts, and describe the organization and structure of domain knowledge clearly. The establishment of the ontology model provides a series of criterions with well defined and formal conceptual for the specific application of tourism emergencies, enhance the capacity of acquisition and representation of the domain knowledge, reuse and share the ontology easily, so as to provide a unified semantic foundation and framework model for the occurrence, development and evolution mechanism of events.(2) An event framework model is established based on the domain knowledge ontology. Combined with the attribute reduction method based the rough set, they are used together to extract the feature of the Web document for tourism emergency. Attribute reduction is completed for attributes and individuals contained in the domain ontology by the rough set method to obtain attributions with larger feature contribution. Then, establish a framework model of the event according to the ontology knowledge, propose a feature extraction method called DK-CHI that extract features based on domain knowledge and CHI statistics. Compared with other feature extraction method based on the objective function, this method can greatly increase the efficiency of feature extraction.(3) An event state evaluation method is proposed for tourism emergency, which uses Internet information as evaluation parameters to establish a state evaluation system and an evaluation function. A classification method of the time-sequence Web document called TS-ISVM used to track topic is proposed, which can realize the recognition and acquisition of follow-up reports for current topics. Experiments show that TS-ISVM classification algorithm can get good classification results for small training sets, improve the training speed without significant accuracy reduction, and get good results of topic tracking. A state evaluation system considering the hot pages, page numbers, the time decay rate and the topic similarity is established, and assesses the event state according to the state evaluation function. Experiment proves that our evaluation model is reasonable, and can reflect the occurrence and development of tourism emergency well.(4) An evaluation method for evolution relation is presented for tourism emergency, which is based on topic clustering with the same topic of events, to assess and predict the evolution relation by the event property distance. Specifically speaking, in order to study the evolution relation between different sub-events of tourism emergency, we propose the incremental clustering algorithm called EGIC, based on the Gauss density and EFD distance, used for clustering in the Internet topics. This method classifies relevant Web documents into different topics and clusters incrementally for the time-sequence Web documents.Experiment shows that EGIC is valid for clustering and detection of topics. Furthermore, the clustering topic can correspond to the realistic topic well. An evaluation method of topic evolution relation called TERE is proposed and a probabilistic model between event property distance and topic evolution relation is also created. It is able to predict the relationship between two topics according to the distribution of event property distance, and our results prove the effectiveness of our TERE method.