IEEE Computer Society Newsfeed

Subscribe RSS

« Back

New Tool Combs Scientific Texts to Find Facts, Create Hypotheses

UK researchers are developing a computer tool able to read scientific literature, make connections between facts, and develop hypotheses. Particularly in the medical field, the rate of publication is expanding exponentially, making it difficult to track all research results in a specialty. The new system—which the University of Cambridge’s Natural Language and Information Processing Group developed—could help scientists find pertinent articles relevant to their needs and tell them why the information is important. The Cambridge researchers are developing natural-language text mining specifically for biomedical literature related to the cancer risk assessment of chemicals. Although there are procedures in place to determine what the relationship between chemical exposure and the probability of developing cancer is, the process—particularly conducting a literature review manually—is lengthy. With researchers from Sweden’s Karolinska Institutet, the UK scientists created a tool called CRAB that can automate the literature review. The system trawls MEDLINE—a database of life-sciences and biomedical information—for relevant articles, looks at their abstracts and creates a profile of relevant material about the specific chemical. Researchers also use the tool to explore new hypotheses based on the information in the literature. The researchers plan to make CRAB available online while continuing to improve text-mining techniques. ( UK)(University of Cambridge)(CRAB)

Trackback URL: