Translation by text categorisation: medical image retrieval in ImageCLEFmed 2006

Authors:
Julien Gobeill;Henning Müller;Patrick Ruch
Affiliations:
Medical Informatics, University and Hospitals of Geneva, Switzerland;Medical Informatics, University and Hospitals of Geneva, Switzerland;Medical Informatics, University and Hospitals of Geneva, Switzerland
Venue:
CLEF'06 Proceedings of the 7th international conference on Cross-Language Evaluation Forum: evaluation of multilingual and multi-modal information retrieval
Year:
2006

Citing 7
Cited 0

Pivoted document length normalization

SIGIR '96 Proceedings of the 19th annual international ACM SIGIR conference on Research and development in information retrieval
Combining classifiers in text categorization

SIGIR '96 Proceedings of the 19th annual international ACM SIGIR conference on Research and development in information retrieval
Content-based query of image databases: inspirations from text retrieval

Pattern Recognition Letters - Selected papers from the 11th scandinavian conference on image analysis
Automatic assignment of biomedical categories: toward a generic approach

Bioinformatics
Query translation by text categorization

COLING '04 Proceedings of the 20th international conference on Computational Linguistics
Term proximity scoring for keyword-based retrieval systems

ECIR'03 Proceedings of the 25th European conference on IR research
Overview of the ImageCLEFmed 2006 medical retrieval and medical annotation tasks

CLEF'06 Proceedings of the 7th international conference on Cross-Language Evaluation Forum: evaluation of multilingual and multi-modal information retrieval

Quantified Score

Hi-index	0.00

Visualization

Abstract

We present the fusion of simple retrieval strategies with thesaural resources to perform document and query translation by text categorisation for cross-language retrieval in a collection of medical images with case notes. The collection includes documents in French, English and German. The fusion of visual and textual content is also treated. Unlike most automatic categorisation systems our approach can be applied with any controlled vocabulary and does not require training data. For the experiments we use Medical Subject Headings (MeSH), a terminology maintained by the National Library of Medicine existing in 12 languages. The idea is to annotate every text of the collection (documents and queries) with a set of MeSH terms using our automatic text categoriser. Our results confirm that such an approach is competitive. Simple linear approaches were used to combine text and visual features.