Home

K. Experimentation and Integration Plans (edited by Prueitt and Myers)

Four experiments are planned for Phase 1. Experiments focus on previously analyzed data and on aspects of ontology operations, unique and efficient data constructions (CCM, Cognitive Graphs, I-RIBs and Polylogics) and implementation as a functional extension (revision) of data management systems that work on hash tables or constructions using categoricalAbstraction, and tool integration into a collaborative environment. The collaborative environment, Groove, will be used by engineers from a core group of nine companies and a group of eleven consulting social/cognitive scientists. The larger tests use massive data and include multiple forms of analysts. These tests will demonstrate wide applicability of the knowledge sharing core concept (see section xx). We will present the now separated functionality currently in a significant number of commercial systems within a private virtual network, complete with a primitive distance learning capability and (objective) educational modules developed by leading scholars. The result will be a common licensing on core innovations (see section yy)

Experiment 1: Bench Test 1: Fable texts have reduced complexity in vocabulary allowing for focus on discourse-level analysis, and conceptual representation. Prior studies of “fable thematic representation” will be compared and extended. Operational Hypothesis: Fable passages can be extracted and grouped according to conceptual schema. Collaborative Hypothesis: A diverse group of social/cognitive scientists and computer scientists can collaborate in the development and deployment of conceptual rollup tools based on a transparency to the patents that are core to many commercial systems.

Experiment 2: Bench Test 2: The relationships among a small set of patent disclosures in the knowledge science field were mapped. Hypothesis: Differential ontology and supportive technologies are able to delineate small and large differences in concept types, and generate systemic networks of meaning. This map can be used to predict new forms of innovations as expressed in newly filed patent applications.

Experiment 3: Simulation Test: We have a large corpus containing unknowns that require interpretation by analysts. Past work is focused on extraction of themes and ‘story lines’, and how those themes change and new variants spread in response to salient events. One of our preliminary data set is composed of web harvests over an 8-month period from Islamic social groups in 14 countries. Hypothesis: Islamic scholars and social/cognitive scientists can act together as intelligence analysts in a re-analysis of this data set using the new knowledge technology tools. A new data set is to be developed based on a significant web harvesting process that will be experimental within our group.

Experiment 4: Application 1: Monitor medical literature for changing bio-defense concepts. The Virginia BioInformatics Institute (Virginia Tech) will make available part of their large archive and ongoing analysis. Hypothesis: Multiple tools can be added to the basic design and increase fidelity when used in conjunction.

Integration. Formative ontology is convertible to Topic Maps, which has a high fidelity link point to the Cyc reasoning system. Whatever the terminology used, any need to establish interoperable systems can be reconciled and automatically synchronized by SchemaLogic’s tools.

Section yy

First Tier knowledge technologies

· NdCore Conceptual Rollup– from Applied Technical Systems;

· SchemaServer from SchemaLogic,

· Semio concept maps from Entrieva,

· NLP++, a system to produce situated NLP processing from Text Analyst International Corporation,

· Probabilistic Latent Semantic Indexing from Recommind,

· Latent Semantic Indexing from SAIC,

· ClearResearch, an automated system for developing parsing rules from ClearForest,

· Polylogics from Pilesys Inc,

· Differential and Formative Ontology from OntologyStream

While our main use of these tools is internal to the project, broad interoperation is an essential feature of the design, and this functional can be readily activated to bridge to other basic innovations. However, the integration of these innovations has to be protected by patent or copyright so that there is NO proprietary intellectual property that cannot be publicly disclosed. Distance learning modules provide a sufficient avenue for any analyst to learn exactly how the specific innovation works.

Groove will be used to support the integration of mental models among analysts and cognitive/social scientists. This is not a trivial point. The analysts/scientists are integral to the complex reasoning system, hence discussions need to be open among them, and with those who are using other means, within a rich and readily used context. Polylogics, situational logics (formative ontology), and metaphoric reasoning using cognitive graphs will rely on human reification at every analytical step.