A platform to automatically retrieve and integrate knowledge relevant to a disease based on a semantic graph representation.
Nentidis, A., Bougiatiotis, K., Krithara, A., & Paliouras, G. (2019). Semantic integration of disease-specific knowledge, 1–8. Retrieved from http://arxiv.org/abs/1912.08633
The iASiS Open Data Graph platform is structured into two individual parts, available in the repositories referenced below :
Harvesting of Biomedical Knowledge: Biomedical Harvesters
- Online retrieval of available up-to-date biomedical literature from PubMed and PMC
- Output: The natural language text of the articles harvested (i.e. abstract or full-text) and relations of articles with their topic areas (i.e. MeSH headings)
- Preprocessing of structured biomedical resources like OBO ontologies and DrugBank
- Input: The files of the resources to be harvested
- Output: Relations between biomedical entities from structured resources (e.g. "concept a" is a "concept b" or "drug a" interacts with "drug b" )
- This step produces data sets in a simple JSON format stored in individual MongoDB collections
Semantic Enrichment of Biomedical Knowledge: Medknow
- Extraction of knowledge from literature natural language text
- Input: The MongoDB collection with natural language text of articles
- Output: Concepts occurring in article text and relations between occurring concepts
- Semantic indexing of harvested relations in a common framework as a semantic graph
- Input: The MongoDB collection of relations between entities (including relations of articles with their topic areas and relations from structured resources)
- Output: Relations between entities represented in a common UMLS-based semantic framework.
- This step produces an integrated knowledge graph in a Neo4j database where knowledge from all harvested resources is uniformly available through graph queries.