MedlineRanker

MedlineRanker ranks Medline abstracts for user-defined biomedical topics by identifying discriminative words from a provided set of abstracts and scoring other abstracts to prioritize relevant literature.

Key Features:

• Automated Discriminative Word Identification: Identifies words significantly enriched in a user-provided set of Medline abstracts compared to a random sample from the Medline database.
• Relevance Scoring and Ranking: Scores and ranks other Medline abstracts based on the presence and weights of the identified discriminative words.
• Inclusion of Recent Publications: Ranks recent publications that have not yet been indexed with specific keywords to surface emerging research.
• Efficiency and Scalability: Processes and ranks millions of Medline abstracts to support large-scale literature retrieval.

Scientific Applications:

• Systematic reviews: Prioritizes relevant Medline abstracts to support literature selection in systematic reviews.
• Meta-analyses: Facilitates identification of candidate studies for inclusion in meta-analyses by ranking relevant abstracts.
• Hypothesis generation: Supports exploratory research and hypothesis generation by highlighting topic-relevant literature.
• Literature surveillance: Enables detection of recent developments and trends by ranking publications not yet indexed with keywords.

Methodology:

Text-mining extracts and analyzes key terms from a user-provided set of abstracts, compares term frequencies against a broader Medline corpus to identify discriminative words, and uses those words to score and rank other Medline abstracts.