![]() |
![]() |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
![]() ![]() ![]() ![]() ![]() |
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
![]() |
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Home | Research | Research Programs | Information Analysis and Retrieval |
Information Analysis and RetrievalEurekaSeekThe NRC Institute for Information Technology (NRC-IIT) research project EurekaSeek, which began in July 2001, is a supplementary project to NRC-IIT’s ongoing work in LitMiner. EurekaSeek is an automated knowledge-discovery tool that uses the principles of linked-literature analysis in text-mining a database. The particular database examined in this project is PubMed, the database of citations and abstracts to biomedical and other life science journal literature. The EurekaSeek project aims to discover research opportunities by identifying logically related subject headings in the PubMed database that do not co-occur in the literature. Previous work had focussed on relationships between words in the text and/or titles of the articles themselves. EurekaSeek explores a more efficient technique, whereby querying on Pubmed subject headings alone, the essence of linked-literature analysis can be retained without the need to parse raw text. NRC-IIT researchers are investigating whether EurekaSeek can be used to predict new co-occurrences between medical subject headings. While the proportion of identified linkages generated is still too small for the process to be a practical aid to research, it may be possible to increase the precision of EurekaSeek and extend its applicability to other databases. Research ContactJeffrey Demaine Business ContactRandall Milburn |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
![]() |
|