Overview of IO Informatics US Patents #6,988,109, #7,702,639, #12/758,415
Data structuring object and algorithms
Object-oriented multi-parametric normalization algorithms efficiently make distributed, multi-method data comparable. This makes it possible to gather "actionable" knowledge by associating, comparing, and defining relationships between previously incomparable data.
The company’s VSS (vector subset selection) algorithm allows users to define and access proper subsets within any dataset - without programming. Together with normalization methods, this allows users to select any data or subset of data for user-driven or automated and iterative analysis, transformation, and comparison.
Methods for handling ontologies, including ontology import, semantic web capability (RDF, OWL, GO, DAML, etc) are also described. Additional "semantic web" oriented methods are defined for relationship detection, dimension reduction and learning functions. These are made possible through the interaction of IO Informatics' use of object ontologies, interacting with proprietary subset-based data structuring methods, query automation, results training and weighted results linking.
back to top
Human / computer interaction
Point-and-click / WYSIWG data structuring methods allow users to "point" to select subsets of data to select, transform, and compare previously undefined data.
Sentient takes previously incompatible data
into a common framework that
allows users to interact directly with the data. Data associations are made easily.
This gives end users convenient, unified access to previously disconnected data, with the ability to define data fields and to detect and define associations and relationships – in a "point and click" manner - without programming.
Additional disclosure includes the description of mobile / virtual data record, distributed database functions. This supports the distribution of objects with auditing, byte-level security, and reporting and database functionality. User-interactive curation methods support promotion and distribution of ontologies, as well as methods to define workspaces and links as database fields, and to weigh links between objects – without programming.
back to top
Visual Creation and Deployment of Semantically Integrated and / or Federated Queries
Novel methods, workflows and user interfaces for creating a semantic search pattern are described allowing end users to pick from or otherwise create a visual network of linked data that can be modified and applied as a search pattern.
This allows searching across semantically integrated databases or across semantically federated data sources. Visual SPARQL queries are directly generated automatically from interactive, user selected sub-networks without requiring knowledge of the SPARQL query language or any further user action.
Users can visually select elements or pattern from a network, and Sentient transforms it into a semantic query without need for programmatic intervention.
Visual SPARQL capabilities have been extended to include filters on ranges, weighting, inclusion and exclusion criteria, etc. Sets of such SPARQL queries may be captured and saved as arrays. Patterns or arrays of patterns can be iteratively refined with new confirmed datasets to increase prediction precision, and can include one or multiple scoring algorithms for decision support.
back to top