Project title: "Optimization of the MEDPILOT search functions: Multilingualism and normalization of linguistic variants"
Product description and objective
The aim of this project was to improve the search functions and results provided by the MEDPILOT search portal. The results of this project have also been incorporated into the new search portal LIVIVO, that has replaced MEDPILOT in 2015.
Computational linguistics technologies (Morphosaurus-technology) were used to significantly improve the quality of search results.
Special search and classification solutions can help users run very fast, user-friendly search queries in medical texts and databases. The texts are subjected to an intelligent semantic analysis which brings together relevant contents and makes them easier to find. This improves the search experience for users and increases the number of search results. The use of optimised search queries makes the entire search process more efficient.
Using traditional search technology, users get a different number of hits when they search with the singular or plural form of a search term, even though it would make more sense to present them with the same number of search results (i.e. by not differentiating between the syntactic variants). Another issue is that students in their first year at university are likely to use different search terms than experienced medical practitioners. By implementing this new search technology, it should be possible to present both groups with the same number and quality of search results even when they use different search terms. This was achieved by incorporating an integrated thesaurus into the system which enables it to identify when the search terms entered by experts and first-year students are semantically the “same” query.
The role played by ZB MED
ZB MED's task was to evaluate the proposed improvements and to implement the quality assurance measures. Systematic feedback provided to the developers in the form of error and relevance analyses helped to continuously improve the quality of the search results. Equally, retrieval tests and empirical studies incorporating real-life test subjects (medical practitioners and students) were used to determine whether the quality of the search results has improved from a user's perspective. At the same time, usability tests were employed to improve how the search engine functions and how the results are displayed.
The first step involved tailoring the CC MED database – which includes more than 465.000 titles – to the new system. The project was then expanded to include the MEDLINE database, which contains approximately 17 million entries.
Morphosaurus is a joint venture of ZB MED and the company averbis GmbH, a spin-off of the Department of Medical Biometry and Statistics of The University of Freiburg.
Funding for this project was secured within the scope of the Pact for Research and Innovation.
June 2007 – December 31, 2008
- Optimizing and evaluating the MEDPILOT search engine. Boosting medical information retrieval by using a morpheme thesaurus. - By Waldemar Dzeyk and Kornél Markó
In: Journal of EAHIL, Vol. 4, Nr. 1, 2008, S. 14-19
- Effektiv und nutzerfreundlich: Einsatz von semantischen Technologien und Usability-Methoden zur Verbesserung der medizinischen Literatursuche
Author: Waldemar Dzeyk
Cologne, ZB MED, German National Library of Medicine, 2010; 163 pages