Citizen-Centric Access to E-Government Information Through Dynamic Taxonomies

Citizen-Centric Access to E-Government Information Through Dynamic Taxonomies

Giovanni M. Sacco (Università di Torino, Italy)
Copyright: © 2013 |Pages: 12
DOI: 10.4018/978-1-4666-3640-8.ch003
OnDemand PDF Download:
$37.50

Abstract

This chapter focuses on dynamic taxonomies, a semantic model for the transparent, guided, user-centric exploration of complex information bases. Although this model has an extremely wide application range, it is especially interesting in the context of e-government because it provides a single framework for the access and exploration of all e-government information and, differently from mainstream research, is citizen-centric, i.e., intended for the direct use of end-users rather than for programmatic or agent-mediated access. This chapter provides an example of interaction and discusses the application of the model to many diverse e-government areas, going from e-services to disaster planning and risk mitigation.
Chapter Preview
Top

Background

Public information is usually managed by four retrieval techniques, which are frequently used at the same time for different subsets of the information base: a) information retrieval (IR) techniques (van Rijsbergen, 1979) recently called search engines; b) queries on structured databases; c) hypertext/hypermedia links and d) traditional taxonomies.

IR techniques are the obvious choice for laws and regulations, since they are essentially textual in nature. However, their limitations, especially in the legal domain, are well known: Blair and Maron (1985) reported that only 20% of relevant documents in a legal database were actually retrieved. Such a significant loss of information is due to the extremely wide semantic gap between the user model (concepts) and the model used by commercial retrieval systems (words). Other problems include poor user interaction because the user has to formulate his query with no or very little assistance, and no exploration capabilities since results are presented as a flat list with no systematic organization. These latter limitations have been addressed recently. Google and other search engines suggest additional query terms while the user is typing as well as the autocompletion and spelling checking of query terms. Clustering techniques are used to support some sort of exploration, by clustering the documents retrieved by an IR query according to “significant” terms or phrases that occur in them. This approach provides a summary for query results and has been used for instance in the US government portal, firstgov.gov. Cluster summaries do not address the semantic problems inherent in IR and do not increase the recall, which is the critical performance indicator in this context. Rather, they increase the precision of the result because they allow users to quickly skip clusters that are not relevant. In addition, the exploratory capabilities offered by text clustering are quite limited (Sacco, 2000; Hearst, 2006).

Complete Chapter List

Search this Book:
Reset