CAP-KMD aims to provide upstream support activities in axes 1, 2 and 3 of GDFF by publishing exhaustive and quality knowledge on microbe adaptation and biodiversity. CAP-KMD is built upon Omnicrobe, an open integrative database that exposes more than a million facts about microorganisms, their habitats, and their phenotypes. The Omnicrobe content is extracted from reference databases, collection catalogs, and scientific literature using Natural Language Processing methods. The content navigation and search is made powerful by full links to reference taxonomies and ontologies.
The objective of CAP-KMD is to embed quality and confidence assessment in facts provided by Omnicrobe, and then to improve the content quality by employing state-of-the-art text-mining methods. Furthermore the vast amount of data contained in Omnicrobe is an opportunity to further quality and confidence assessment. CAP-KMD will apply formal logic techniques and expert knowledge to automatically detect contradicting and outlier facts in order to improve quality assessment.
The multidisciplinary consortium has identified specific challenges to overcome about extraction of strain names and food-related habitats and phenotypes. The expected outcome of CAP-KMD is a quality qualified massive database on microbe adaptation and biodiversity
knowledge. Combined with -Omics databases, this knowledge will sustain upstream analyses for selecting strains, understanding microbial consortia and designing new ones.