LifeSKIM: Application for large scale biomedical
                  semantic annotations

                Vassil Momtchev1, Georgi Georgiev1, Deyan Peychev1
                               1
                                 Ontotext AD, Sofia, Bulgaria
                    {vassil.momtchev, georgi.georgiev, deyan.peychev,
                                      }@ontotext.com
                             http://www.ontotext.com/lifeskim/



Abstract

   Data integration and interpretation is a very challenging problem for the
fundamental biomedical research and drug development industry. There is an
emerging need for the development of new tools and applications which to
transparently link the unstructured and semi-structured information to the biological
database. LifeSKIM 0.1 is flexible and scalable application for semantic annotations
generation, indexing and retrieval of biomedical entities and documents. Life
Ontology is OWL-Horst knowledge base which contains information about genes,
gene ontology terms, organisms and diseases. It consists of 100 million RDF
statements generated from public biomedical databases Entrez-Gene, GeneOntology,
SNOMED, NCBI Taxonomy and DrugBank. The application demonstrates a corpus
of over 1.2 million documents annotated against Life Ontology resulting in total
10,884,032 semantic annotations to 1,204,063 different entities. LifeSKIM
demonstrates high scalability and the powerful query capabilities for 1) querying and
navigation of knowledge generated from structured (biological databases) and
unstructured (biomedical document) sources; 2) semantic indexing and retrieval of
documents with respect to an ontology; 3) ontology population and learning of new
types of entities from text e.g., gene, cell line and type, DNA and RNA molecules; 4)
Efficient reasoning against the extracted and structured information, e.g., “type I
programmed cell death” is a specific process of “Apoptosis of neutrophils”; 5) co-
occurrence tracking and ranking of entities. The application provides multiple user
and programming interfaces for access and it is fully compliant with the W3C
standards and recommendations.