Search Bot: Search Intention Based Filtering Using Decision Tree Based Technique

Abstract

Most web search engines use only the search keywords for searching. Due to the ambiguity of semantics and usages of the search keywords, the results are noisy and many of them do not match the user's search goals. This paper presents the design of an intelligent Search Bot, which operates as an agent for a user by simulating the user activity of filtering only the relevant search results. It learns from experience and improves its performance with time. The focus is to obtain a user's search intention or requirement from the search query, and then to deliver results accordingly. The user first trains the system according to his search intention, by doing binary classification of the search results. Training is followed by knowledge representation and extraction, and then reasoning and analyzing the new search results to determine their relevance classification. The technique is based on the construction of decision trees. It also finds application in news searching, information retrieval from databases and spam mail detection.