Research analysts typically need to pose potentially complex questions whose answers require reasoning over disparate data as well as domain-specific expertise and world knowledge. This application framework supports the development of applications enabling users to construct questions, in English, and receive answers and natural language justifications. The framework facilitates connectivity to external data sources that can be used by Cyc in its question answering.
Answering the Tough Questions: Today’s analysts – business, defense, intelligence, political, medical, financial, etc. – must find critical fragments of information in an ever-growing sea of data, and piece them together to meet their research objectives.
- Unlike data bases, the Cyc Analysis Suite lets analysts pose their queries in English rather than SQL – by carrying out a clarifying dialogue if needed.
- Unlike search engines, the Cyc Analysis Suite returns answers and lines of reasoning behind each answer, not just relevant hits, by drawing on federated data sources and human-like common sense reasoning.
Solutions to Your Analysis Challenges: The Cyc Analysis Suite provides tools specifically designed to support complex knowledge-fusion tasks. The tools extend the general Cyc Analytical Platform, customized for specific industry sectors with domain-tailored ontologies, knowledge packs, lexicons, business rules, and interfaces. Cycorp’s Semantic Knowledge Source Integration and question-answering API’s permit further customization to interface with your specific knowledge management and analysis environment.
Answers, Justifications, and Visualizations: The Cyc Analysis Suite’s tools provide a variety of information displays, offering the analyst alternative views of the data, including maps, event timelines, relationship networks, and dynamically-assembled fact sheets about domain-relevant concepts (e.g., companies, terrorist groups, acquisitions, sales, attacks, countries, specific individuals, and so on.) Each answer is accompanied by a detailed, dynamically-generated, narrative-style justification, with citation links to all sources that support (or argue against) that conclusion.