NLIMED

NLIMED translates natural language queries into SPARQL queries to discover and retrieve model entities such as variables, equations, and whole models annotated in RDF-based biosimulation repositories like the Physiome Model Repository (PMR) and the BioModels database.

Key Features:

• Natural Language Processing (NLP): Segments user queries into phrases and annotates them against ontology classes and predicates using specialized NLP techniques.
• Semantic Annotation: Operates over the Resource Description Framework (RDF) as recommended by the COmputational Modeling in BIology NEtwork (COMBINE) to locate variables, equations, and models within biosimulation repositories.
• SPARQL Query Generation: Composes annotated query components into executable SPARQL queries via a SPARQL Composer.
• Indexing and Ranking: Uses an indexing system to rank potential query results based on relevance and accuracy.
• Repository Compatibility: Interfaces with RDF-annotated repositories, explicitly including the Physiome Model Repository (PMR) and the BioModels database.
• Adaptation from Query History: Adapts result prioritization by learning from historical query records to improve relevance.

Scientific Applications:

• Model entity discovery: Locating variables, equations, and model components within biosimulation models for reuse and analysis.
• Semantic search across repositories: Performing ontology-driven searches over RDF-annotated repositories such as PMR and BioModels.
• Mapping queries to ontologies: Translating natural language descriptions into ontology classes and predicates to support reproducibility and interoperability of biosimulation models.

Methodology:

Query chunking and annotation of phrases against ontology classes and predicates; composition of annotated components into SPARQL queries using a SPARQL Composer and an indexing system; ranking of results and adaptation using historical query records.