18 September 2007

Congratulations to Paul Ferguson who successfully defended his thesis and will be awarded the degree of PhD.

The title of Paul's thesis is "Index Ordering by Query-Independent Measures".

He completed his PhD in the Centre for Digital Video Processing (CDVP), Adaptive Information Cluster (AIC) and the School of Computing, DCU under the supervision of Prof. Alan Smeaton.

Brief description of Project:

Rather than simply utilising more computing resources in order to perform retrieval on large text collections, we investigated ways in which to only search a limited amount of the collection at query-time, in order to speed up this retrieval process (as well as allowing retrieval to be carried out with limited computing resources). Although, in doing this we aimed to limit the loss in retrieval efficacy (in terms of accuracy of results).

In order to do this we identified a number of different query-independent measures that can approximate the documents' query-independent quality in the collection, rather than relying solely on a term-weighting approach to determine the importance of a document. In this way we can choose to limit the amount of information to search through, by eliminating the documents of lesser importance, which not only makes the search more efficient, but should also limit any loss in retrieval accuracy.

In our work we applied a number of traditional combination techniques, as well as machine learning approaches to the task of combining these measures together, to provide a more accurate overall measure. We also provided techniques to allow documents to be effectively eliminated from a sorted inverted index while retaining the same level of performance. This research should allow fast and effective retrieval of large collections of documents be carried out much more efficiently than using traditional retrieval techniques, while also retaining the same level of accuracy for a typical web user.

This project was generously funded by Science Foundation Ireland (SFI).

Back to News Headlines