Semantic Web Techniques for Yellow Page Service Providers

Abstract

Applications providing “yellow pages information” for use over the web should ideally be based on structured information. Use of web pages providing unstructured information poses variety of problems to the user, such as use of arbitrary formats, unsuitability for machine processing and likely incompleteness of information. Structured data alleviates these problems but we require more. Capturing the semantics of a domain in the form of an ontology is necessary to ensure that unforeseen application can easily be created at a later date. Very often yellow page systems are implemented using a centralized database. In some cases, human intermediaries accessible over the phone network examine a centralized database and use their reasoning ability to deal with the user’s need for information. Centralized operation and considerable central administration make these systems expensive to operate. Scaling up such systems is difficult. They behave like isolated systems and it is common for such systems to be highly domain specific, for instance systems dealing with accommodation and travel. This paper explores an alternative – a highly distributed system design meeting a variety of needs – considerably reducing efforts required at a central organization, enabling large numbers of vendors to enter information about their own products and services, enabling end-users to contribute information such as their own ratings, using an ontology to describe each domain of application in a flexible manner for uses foreseen and unforeseen, enabling distributed search and mashups, use of vendor independent standards, using reasoning to find the best matches to a given query, geospatial reasoning and a simple, interactive, mobile application/interface. We view this design as one in which vendors and end-users do the bulk of the work in building large distributed collections of information in a Web 2.0 style. We give importance to geo-spatial information and mobile applications because of the very wide-spread use of mobile phones and their inherent ability to provide some information about the current location of the user. We have created a prototype using the Jena Toolkit and geo-spatial extensions to SPARQL. We use simple and shallow reasoning to give inferred information in addition to explicitly stored information. We have tested this prototype by asking a group of typical users to use it and to provide structured feedback. We have summarized this feedback in the paper. We believe that the technology can be applied in many contexts in addition to yellow page systems. The essential features are the involvement of a set of creators of information and a set of end-users, the two sets not being mutually exclusive and the use of an ontology that can span a number of domains in the context of product and service vendors.