October 12, 2010
CALIT2-affiliated computer science professors Chen Li and Xiaohui Xie have been awarded a $371,000 two-year grant from the National Institute of Health to develop novel methods for searching biomedical literature.
The MEDLINE database, complied by the United States National Library of Medicine, is a comprehensive bibliographic database of life sciences and biomedical information. The database contains more than 19 million records from approximately 5,000 selected publications, covering much of the literature in biomedicine and health. The database is also growing fast, with thousands of updates every day.
Searching MEDLINE has become an indispensable component in the daily life of medical practitioners, biological researchers and an increasing number of patients who prefer to seek medical information on their own.
Currently, searching MEDLINE is primarily conducted through the PubMed Web server, which is maintained by the NIH’s National Center for Biotechnology Information and handles millions of searches per day. Given the high popularity and importance of this database, Li believes it is critical to study how to make MEDLINE search more powerful and easier to use.
Li and Xie, together with collaborators at Tsinghua University, China, have developed a system called iPubMed (http://ipubmed.ics.uci.edu), to study how to support instant, error-tolerant search on MEDLINE publications.
Their published paper in the journal Bioinformatics (http://bioinformatics.oxfordjournals.org/content/26/18/2321.abstract) demonstrates that the experience of searching MEDLINE can be significantly improved by instant search.
In the new search paradigm, a user can view the search results instantly as he or she types each letter of the query. Because the user can modify the query “on the fly” according to the instantly returned results, it can take much less time to locate the right items.
This new search model is also gaining popularity in other domains. For instance, Google recently released a new web search tool called Google Instant, which implements a similar idea. The error-tolerant feature is especially important when the user does not remember the exact spelling of keywords, such as disease names or author names.
The NIH grant will support Li and Xie’s research efforts to further improve the iPubMed system.
Li’s research interests are in the fields of database and information systems, including Web search, large-scale data management, data cleansing and data integration.
Xie’s research focuses in machine learning, bioinformatics, computational biology and neural computation. He is interested in both developing novel machine learning theory and algorithms, and applying them to practical problems, such as biology and medical science.
— Sherry Main, Bren School of ICS
