Browsing by Subject "MEDLINE"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    A Case Study for Massive Text Mining: K Nearest Neighbor Algorithm on PubMed data
    (Office of the Vice Chancellor for Research, 2015-04-17) Do, Nhan; Dundar, Murat
    US National Library of Medicine (NLM) has a huge collections of millions of books, journals, and other publications relating to medical domain. NLM creates the database called MEDLINE to store and link the citations to the publications. This database allows the researchers and students to access and find medical articles easily. The public can search on MEDLINE using a database called PubMed. When the new PubMed documents become available online, the curators have to manually decide the labels for them. The process is tedious and time-consuming because there are more than 27,149 descriptor (MeSH terms). Although the curators are already using a system called MTI for MeSH terms suggestion, the performance needs to be improved. This research explores the usage of text classification to annotate new PubMed document automatically, efficiently, and with reasonable accuracy. The data is gathered from BioASQ Contest, which contains 4 millions of abstracts. The research process includes preprocess the data, reduce the feature space, classify and evaluate the result. We focus on the K nearest neighbor algorithm in this case study.
  • Thumbnail Image
    Item
    IN Health Connect: Connecting Local Health Services to Quality-Filtered Health Information
    (H.W. Wilson Company, 2005) Richwine, Peggy
    To many librarians, the term MEDLINE has connotations of a huge, complex database that returns far too many citations with little relevance or readability for most library users. And although some might expect that MedlinePlus is more of the same, it is really QFWBFTCHI – quality-filter, web-based, full-text, consumer health information. Unlike MEDLINE, MedlinePlus is relevant and readable for the library user seeking health information. Librarians in Indiana have contributed to a unique dimension of MedlinePlus, IN Health Connect, which offers state residents consumer health information specific to the region where they live. Some background on MedlinePlus prefaces the development of this initiative.