BioTAGME

BioTAGME integrates TAGME and DT-Hybrid to extract and model relationships among biological entities from PubMed titles and abstracts, producing a Knowledge Graph for literature-based biomedical knowledge discovery.

Key Features:

• TAGME entity annotation: Uses a Wikipedia corpus to annotate entities and phrases extracted from text.
• DT-Hybrid network inference: Applies a network-based inference methodology to predict relationships among entities.
• Knowledge Graph construction: Models relationships among biological terms and phrases extracted from titles and abstracts.
• PubMed text processing: Operates on titles and abstracts of papers available in PubMed as input data.
• Scala + Spark backend: Implements the back-end in Scala and distributes computation across a Spark cluster.
• Neo4j graph storage: Stores and enables querying of the generated Knowledge Graph in Neo4j.

Scientific Applications:

• Hypothesis generation: Infers novel knowledge and testable hypotheses from biomedical literature.
• Relationship discovery: Identifies and ranks relationships among biological entities and phrases.
• Literature-based knowledge discovery: Builds comprehensive biomedical Knowledge Graphs to support research-driven insight extraction.

Methodology:

Combines TAGME entity annotation (Wikipedia corpus) with DT-Hybrid network-based inference to create a Knowledge Graph from PubMed titles and abstracts; backend implemented in Scala using a Spark cluster; graph data stored and queried in Neo4j.