The anatomy of a domain-specific search-phrase suggestion tool for literature digital libraries

Abstract EN

Suggesting search terms in keyword-based search interfaces is becoming common place.

Google Suggester, for instance, in a history-based one as it provides suggestions to queries using a dynamic list of pre-saved popular queries.

However, this approach suffers from accuracy and efficiency problems.

In this paper, the anatomy of a large-scale efficient suggester for literature digital libraries is proposed.

This suggester makes suggestion from the pre-analyzed document collection to be searched.

The proposed suggester promises providing human-like suggestions that are automatically extracted from the repository of web documents, i.

e.

publication in the context of digital libraries web pages in the context of the web.

The proposed tool also provides an accurate term ranker to assign importance scores to suggested keywords within the contexts where they are observed.

The proposed tool promises a more scalable and user-friendly search-keyword suggester when compared to its history-based competitors.