A Personalized Search Framework for Industrial Safety and Health Information Retrieval

摘要

The emergence of the WWW brought about new searching and querying difficulties. It is evident that the Internet and its most popular service WWW have changed our everyday lives. Normally the search engines use the keyword based querying methods to retrieve the web documents as a result. But the fact is most of the results retrieved are not relevant to the users, because of the contextual ambiguities. A programmer may search for the query "SAFETY", referring to the state of being safe in various context. While it is searched by a home maker, it refers to the home safety. If it is searched by an expert who is training the people in safety and health management, his need may be of different nature. In order to produce the result based on the context, Personalization Methods are introduced. In the proposed system, personalization is done in two phases, (ⅰ) Building User profiles (ⅱ) Reranking the SERPs (Search engine Result Pages). The browsing behavior of the user is represented in the form of user profile which consists of static initial information and dynamic search history of the user. Ranking algorithm takes the both static and dynamic factors weight as input for personalized reranking operation. The degree of personalization is also measured using the Jaccard Co-efficient and Hamming Distance. A Safety and Health Management System (SHMS) is a systematic approach that manages safety and health activities. It is covering occupational safety and health programs, policies, and objectives into organizational policies and Procedures. The dataset from the SHMS domain is used for the experiment and the profiles of different types are created. It shows the percentage of improvement in the relevancy of search results under various conditions in terms of precision and recall.