Information Access through Search Engines and Digital Libraries

Citation

Smith, A.G. (2009), "Information Access through Search Engines and Digital Libraries", The Electronic Library, Vol. 27 No. 1, pp. 187-188. https://doi.org/10.1108/02640470910934867

It is easy to think that information retrieval research for the web takes place in the proprietary environment of search engine companies. This ignores the fact that the currently most popular search engine, Google, emerged from an academic research project. So the search systems that we will use in future are probably being developed now in a university research centre. The University of Padua (which in the past has hosted indefatigable researchers such as Copernicus, Galileo and Casanova) includes the Information Management Systems group from which this collection of papers derives.

Agosti's introductory paper provides an interesting historical overview of IR, looking particularly at the role of the user in guiding research. She gives a number of examples where librarians and users have contributed to the development of IR systems, for example in the Digital Archive of Venetian Music (ADMV).

An important research area at Padua is automatic hypertext linking, and Crestani looks at how the experience of developing a hypertext textbook was applied to the prioritisation of suspects in a criminal investigation, for example mining text descriptions of crime scenes, geographical information, and social networks. The privacy implications of these techniques are not discussed.

Relevance has been a long‐standing research topic in information retrieval, and Mellucci investigates how the context of the information need, which is not taken into account by existing search systems, affects the search. This paper explores how context could be modelled to produce a “context‐aware” search engine that uses automatically generated links.

As might be expected, in a European research context multilingual searching is an issue, and Mellucci and Orio look at how stemming algorithms with little linguistic knowledge can be used to search in a range of European languages. The linguistic theme is also explored by Crivellari, Di Nunzio and Ferro in their paper on the evaluation of multilingual retrieval systems.

With better bandwidth, visualisation of search results is a developing area, and Di Nunzio investigates this in his paper on 2DPM – a two dimensional probabilistic model – that will help categorise and visualise text sets in digital libraries.

Ranking by link analysis has been the magic bullet that has made Google so successful, and Pretto's paper reviews the mathematical theory of link analysis, which has been applied in many different fields apart from IR.

In an increasingly interactive web, the concept of annotation of documents is developing in new ways. Ferro explores how annotations have been used, from classical times to the semantic web. He then proposes a formal model of annotation, and how this can be applied to IR algorithms.

Music is a problem in multimedia digital libraries because end‐users have a wide range of knowledge yet may not be able to utilise music‐related metadata. Orio suggests using the methods of approximate indexing to match music documents to requests, and tests the concept on a collection of Europop.

While this work is closer to a journal issue than a book in terms of number and range of articles, and the approach may be more mathematical than many library professionals are used to, it is rewarding to dip into, and may provide some insights into the shape of future search tools and digital libraries.