Is Open Cyc doomed to be the new
Is Open. Cyc doomed to be the new Esperanto, or is OOR doomed to be the new Electronic Data Interchange, or -- even worse -- both! Doug Lenat Cycorp Our content What we’d want a good host to provide Given the other, funded, open ontology repository projects going on in the world (e. g. OKKAM), does it need one more?
Our Content Open. Cyc (www. opencyc. org) The Cyc Ontology made 100% freely available (yes, 100% free even for commercial purposes) Available for download on Source. Forge Over 30, 000 “users” Research. Cyc (researchcyc. com) Open. Cyc + millions of hand-engineered assertions Free for R&D purposes Current users: 300 research groups (1/2 academic) 2
What are people doing with it? • USAF 45 th Space Wing: Decision Support • USNavy: Threat Scenario Detection • US Forest Service: Regulatory Compliance • Lar. KC: Large Knowledge Collider • Medical Research Center: Clinical Trial Cohort Selection (doctors can now directly formulate complex FOPC queries via interactive clarification dialogue; DBs) • Glaxo: semi-automatic ontology alignment across multiple large domain-specific info sources 3
What’s in Open. Cyc Explicitly: 300 k terms; 14 k predicates; 57 k classes; 2 million assertions; infin. more nonatomic terms and inferred assertions (#$isa 596215) (#$genls 99198) (#$disjoint. With 6114) (#$result. Isa 4277) (#$result. Genl 1206) (#$arg. Isa 35617 (#$arg. Genl 5398) (#$arg 1 Isa 16748) (#$arg 1 Genl 2354) (#$arg 2 Isa 14114 (#$arg 2 Genl 2283) (#$arg 3 Isa 3486) (#$arg. Format 5493) (#$arg 2 Format 3320) (#$functional. In. Args 1427) (#$arity 16416) (#$arity. Min 958) (#$comment 57305) (#$genl. Preds 7440) (#$negation. Inverse 990) (#$genl. Mt 26078) (#$denotation. In. English 409745) (#$synonymous. External. Concept 13916) 4
Systems and Processes event. Occurs. At resource conveyer energy source ‘lifetime’ of system provider. Of. Motive. Force resource synthesizer done. By boundary transporter 5
Specializations Ecosystem agent. In. Ecosystem Functional. System component. In. System Organization has. Members Organism anatomical. Parts Ecological. Process Autocatalytic. Process Culture-Practice Metabolism 6
Ecosystem Classes Ecosystem Aquatic Life Zone genls Biome genls Desert Ecosystem Tropical Rainforest Ecosystem Chaparral Ecosystem Tundra Ecosystem Taiga Ecosystem Grassland Ecosystem 7
Ecosystem genls Chaparral Ecosystem climate. Of. Ecosystem. Type terrain. Climate. Type Geographical Region Mediterranean Scrub Mediterranean Climate Cycle has. Climate. Type Territory Of Santa Barbara, CA genls 8
What We’d Want a Good Host to Provide A commitment to use – to have contributors all provide content under – some Creative Commons license, as opposed to e. g. a GNU license Retention of the provenance/lineage of contributed ontological content Agreement on some of the most fundamental ontological relations Agreement on a small set of inter-ontology alignment relations 9
Given the other, funded, open ontology repository projects going on in the world (e. g. OKKAM), does it need one more? OKKAM is already a funded UE FP 7 project (~$10 M, 3 -years) that started 2 months ago. Ontologizing individuals (including organizations such as the USArmy and IBM as individuals), providing a unique identifier and agreed-on set of properties for each individual DBpedia extracted the content of fact boxes from Wikipedia + 35 opensource ontologies; KBpedia EU STREP ($3 M) follow-on and will include true ontology-merging Lots of other projects which other speakers in this panel will no doubt mention 10
…and, coming to a lab near you in Feb 2008… The Large Knowledge Collider 11
Organisation Country Universität Innsbruk Austria Astra. Zenica AB, R&D Sweden CEFRIEL S. c. r. l. Italy Cycorp, Raziskovanje in Eksperimentalni Razvoj, d. o. o. Slovenia Universität Stuttgart, HPCC Germany Max Plank Gesellshaft Germany Sirma Group, Ontotext Lab Bulgaria Saltlux Korea Siemens Aktiengesellshaft Germany University of Sheffield United Kingdom Vrije Universiteit Amsterdam Netherlands Beijing University of Technology PRC WHO: International Agency for Research on Cancer France FP 7 IP - Lar. KC Consortium 12
Query Method massive distributed incomplete reasoning Problem Method Problem Method Problem zillions of assertions 13
Open Source Platform • Distributed platform will be freely distributable and modifiable Based On Cyc Inference • First version of platform based on streamlined Java build of Cyc Inference Engine Plug-in Experiments • Plug-in architecture allows for experiments in access, reasoning, aggregation… 14
- Slides: 14