Document Search Using Content, Structure and Properties

Abstract

More and more information is nowadays being stored in electronic form. For users it becomes more and more difficult to find information in current document bases. Essentially, there are two different ways for a user to find information in document bases. The first way is to browse. The user starts with a particular document and follows links from that document to other documents until her information need is satisfied. However, the user may also not find the desired information and give up, or become distracted and follow other paths that are considered interesting. The second way is to query the document bases using a search engine. The most common kind of search engine is based on an index that relates keywords to documents. The index is created based on an analysis of the text in the documents. The search is then essentially a table lookup in the index. An example of such a search engine is AltaVista ([AltaVista]). This kind of engines is widely used nowadays. They are easy to use and fast and often give the user at least an approximate answer to the query. Therefore, these systems perform well when a user only needs a crude answer to a query or when a user only needs a starting point from which to browse or query further.