PURE

PURE recommends PubMed articles using content-based filtering and model-based clustering to deliver personalized literature recommendations for researchers.

Key Features:

• Content-Based Filtering: Analyzes topics and keywords from selected articles to prioritize similar PubMed articles.
• Model-Based Clustering: Groups similar articles using model-based clustering to identify patterns within the user's selected literature.
• Predictive Preference Modeling: Builds a predictive model of user preferences from clustered articles to guide recommendation ranking.
• Daily PubMed Updates: Updates the PubMed article database daily to incorporate recent publications into recommendation generation.

Scientific Applications:

• Automated Literature Discovery: Identifies relevant PubMed articles aligned to a researcher's interests, reducing manual search time.
• Keeping Current with Research: Surfaces newly published PubMed articles that match established preferences to help researchers stay up to date.

Methodology:

Ingests user-selected PubMed articles, applies model-based clustering to derive a predictive model of user preferences, uses content-based filtering to score and recommend new PubMed articles, and updates the PubMed database daily.