LitPathExplorer

LitPathExplorer extracts and organizes literature-derived events and assigns confidence scores using text mining and semi-supervised learning to support pathway model exploration and curation.

Key Features:

• Automated Event Extraction: Automatically extracts events (statements) from published literature that are relevant to existing pathway models.
• Confidence Scoring: Assigns a confidence value to each extracted event based on linguistic features and article metadata.
• Semi-Supervised Learning: Applies semi-supervised learning algorithms to refine extraction models, with reported precision improvements from 61–73% to up to 95% with user involvement.
• Interactive Visualization: Visualizes pathway models enriched with literature-derived events to support exploration.
• Precision and Recall Metrics: Reports event extraction precision of 89% and recall of 71%.

Scientific Applications:

• Pathway Model Curation: Facilitates curation and updating of pathway models by supplying literature-derived evidence for model components and interactions.
• Discovery of Novel Events: Identifies novel biological events reported in the literature for potential incorporation into pathway models.
• Confidence-Based Analysis: Enables prioritization of extracted events for curation and review based on assigned confidence scores.

Methodology:

Uses text mining techniques and semi-supervised learning algorithms to extract events from scientific literature and organizes extracted events by confidence level; methodology was validated by quantitative evaluations reporting 89% precision and 71% recall.