Cogito's Graph-based Relationship Analytics solutions are built on innovative directed graph technology. While representational schemas based on mathematical graph theory date back to the same era as relational databases, Cogito is the first company to deliver a commercially viable graph database designed for relationship analytics.

Once data is fused into a network that expresses the relationships among the data elements, analysts are able to see visual representations of connected data, find patterns and discover connections that are obscured by other kinds of information representation.

Data stored in a graph database is represented with nodes and arcs. A node represents a concrete entity. An arc represents a relationship of some type between two nodes. Any node may be linked to any other node with an appropriate relationship. Values (or weights) may be associated with each individual relationship, providing further options for characterizing the nature of the relationship. With nodes and arcs as atomic elements, Cogito is able to define metadata that describe how atomic data elements are interpreted and used. Cogito's basic model—consisting of classes, instances, relationships and attributes—accommodates most common data modeling situations.

Representing information as nodes and arcs provides several advantages:

• The data model more closely parallels the real world, with connections to everything that is immediately related. This preserves relative meaning and context.
• Information from disparate sources such as multiple relational databases is aggregated and fully normalized without replication.
• Queries reveal how information is related, with network visualization, concentrations of connections, traversal across multiple degrees of separation, and pattern detection and analysis.
• Higher retrieval performance and lower data management, maintenance, and update costs.

NETWORK ANALYSIS: COGITO VS. RDBMS

The table above shows how the performance of a graph database and a relational database (MySQL) compares for two different type of network analysis operations:

1. The "Node Expansion" operation involves finding the neighbors of the neighbors of the neighbors of a given start node. Another, simpler, way to say it is, “expand from a start node out 3 degrees, examining all adjacent nodes in the process.” On a graph with 50 arcs per node this operation examines over 120,000 nodes in less than 6 seconds. The RDBMS takes over 32 minutes to complete the same operation.
2. In the second operation, all paths of up to 4 degrees between two nodes are found. With a 200,000 node graph (50 arcs per node) this operation is done in 1.1 seconds vs. 7 minutes using RDBMS technology like MySQL.

These two examples highlight the value of "Graph" vs. RDBMS technology for relationship analytics.

APPLICATIONS

Graph technology and network relationship analysis can be applied to:

• Intelligence and Security - Aggregates large amounts of data from multiple sources to look for hidden connections, predictive patterns and relative context.
• Social Network Analysis - Evaluates connections for clustering, affinities and grouping, pattern interactions, organization structure, link density, subgroup identification, centrality (betweeness, distance, closeness, degree, bridging) and more.
• Manufacturing - Acts as a manifold for stovepipes of information, enabling root cause analysis, dependency analysis, common points/patterns of failure, workflow tracing, and much more.
• Bioinformatics - Connects complex structures stored in different formats with a superior interface to 3D structures, sequences, and alignments.