Medline Ranker

Medline Ranker ranks PubMed abstracts from the Medline database by extracting discriminative words via text mining to score and prioritize literature relevant to a user-defined topic.

Key Features:

• Automated Keyword Deduction: Contrasts word frequencies from a set of relevant abstracts with a random selection to extract discriminative words.
• Relevance Scoring and Ranking: Uses deduced discriminative words to score other PubMed abstracts and produce a ranked list, including recent publications not yet annotated in Medline.
• Efficiency in Processing: Processes millions of abstracts to enable large-scale ranking across the Medline corpus.

Scientific Applications:

• Literature prioritization: Prioritizes PubMed abstracts for literature review and curation across specific and broad biomedical topics.
• Detection of recent publications: Identifies relevant recent publications that lack Medline annotations.

Methodology:

The method accepts an initial set of topic-related PubMed abstracts, contrasts them with a random selection to extract discriminative words using text mining algorithms, applies these words to score and rank other abstracts, and outputs a ranked list.