Fuzzy Information Retrieval

Abstract

Information retrieval used to mean looking through thousands strings texts to find words or symbols that matched a user's query. Today, there are many models that help index and search more effectively so retrieval takes a lot less time. Information retrieval (IR) is often seen as a subfield computer science and shares some modeling, applications, storage applications and techniques, as do other disciplines like artificial intelligence, database management, and parallel computing. This book introduces the topic IR and how it differs from other computer science disciplines. A discussion the history modern IR is briefly presented, and the notation IR as used in this book is defined. The complex notation relevance is discussed. Some applications IR is noted as well since IR has many practical uses today. Using information retrieval with fuzzy logic to search for software terms can help find software components and ultimately help increase the reuse software. This is just one practical application IR that is covered in this book. Some the classical models IR is presented as a contrast to extending the Boolean model. This includes a brief mention the source weights for the various models. In a typical retrieval environment, answers are either yes or no, i.e., on or off. On the other hand, fuzzy logic can bring in a degree of match, vs. a crisp, i.e., strict match. This, too, is looked at and explored in much detail, showing how it can be applied to information retrieval. Fuzzy logic is often times considered a soft computing application and this book explores how IR with fuzzy logic and its membership functions as weights can help indexing, querying, and matching. Since fuzzy set theory and logic is explored in IR systems, the explanation where the fuzz is ensues. The concept relevance feedback, including pseudorelevance feedback is explored for the various models IR. For the extended Boolean model, the use genetic algorithms for relevance feedback is delved into. The concept query expansion is explored using rough set theory. Various term relationships is modeled and presented, and the model extended for fuzzy retrieval. An example using the UMLS terms is also presented. The model is also extended for term relationships beyond synonyms. Finally, this book looks at clustering, both crisp and fuzzy, to see how that can improve retrieval performance. An example is presented to illustrate the concepts.