Process Mining for Entity Association Disambiguation (PMEAD) Navy SBIR FY2010.2 Sol No.: Navy SBIR FY2010.2 Topic No.: N102-176 Topic Title: Process Mining for Entity Association Disambiguation (PMEAD) Proposal No.: N102-176-1296 Firm: Knowledge Based Systems, Inc. 1408 University Drive East College Station, Texas 77840-2335 Contact: Timothy Darr Phone: (979) 260-5274 Web Site: www.kbsi.com Abstract: Knowledge-Based Systems, Inc. (KBSI) proposes to design and demonstrate an innovative "Process Mining for Entity Association Disambiguation" (PMEAD) capability in response to the Navy SBIR solicitation titled "Disambiguation of Entity Association Statements." PMEAD will employ semi-automated provenance generation and process mining to identify rules which can be used to identify redundant and missing statements within entity association RDF graphs. This will improve the quality, fidelity and utility of large-scale RDF graphs derived from real-world data. The PMEAD approach is based on the premise that entities exhibit unique patterns in their behavior and therefore two entities that exhibit the same behavior can be resolved to the same entity with a high degree of certainty. PMEAD automated provenance generation and data mining approach avoids the traditional manually intensive and time-consuming process of developing a rule base for inferring missing relationships and resolving ambiguous entities. Benefits: PMEAD provides the following benefits: (i) avoids the manually intensive and time-consuming process of developing a rule base for inferring missing relationships and resolving ambiguous entities, (ii) automated application of a minimal set of hand-crafted provenance-generation rules to assert provenance statements about an entity association RDF dataset, (iii) process mining for unassisted discovery of process and role models, (iv) automated transformation of the mined process and role models into SPARQL rules, and (v) SPARQL-based inference of entity-disambiguation statements. Return