In Search of a Semantic Book Search Engine on the Web: Are We There Yet?

Abstract

Books being a valuable source of knowledge and learning, have always been searched for on the Web. Traditional Web Information Retrieval (IR) techniques of searching and ranking are applied for this purpose. These techniques, however, are basically designed for dealing with hyperlinked collections of rich text in the form of web pages. Books are inherently different from web pages and the traditional Web IR techniques do not account for their well-organized structure and the logically connected content. Book searching solutions currently available on the Web and in other digital environments, however, do not exploit these implicit semantics resulting in not satisfying the requirements of all stakeholders including readers, authors, publishers, and librarians. These semantics hidden in the well thought out structure and the logical connections in book contents are only visible to human beings. The position put forward here is that most of the available searching solutions treat books as plaintext collections leading to inaccurate and imprecise book search results. Ways and means must, therefore, be found to treat books differently from other web documents and to use their structural semantics and logical connections in the content for searching, ranking and recommendations. Development of comprehensive book structure ontology will help in harvesting these implicit semantics. Similarly, in order to fulfill information needs of the readers, different domain-level ontologies are required so that book contents can be conceptually connected and be made machine ‘understandable’. Moreover, tables in a book consist of structured data and are a rich source of semantics. Similarly, the context of images and figures may be exploited for relating contents within and across books. Discovery and the subsequent utilization of these semantics in book IR process will result in more precise and accurate systems and to the satisfaction of all stakeholders.