Relationship Analysis

Looking at a graphical representation of a document allows a user to decide within seconds whether or not a document is significant, or whether additional analysis or translation is necessary. Below are some examples of link graphs that depict the relationships between entities extracted from a sample set of documents.

Rosoka’s visualization capabilities include the ability to center on a particular entity of interest, crawl out relationships up to 6 degrees, or visualize all relationships between all entities in a corpus. Additionally, Rosoka resolves identities across documents and languages. In other words, if “Rosoka Software, Inc.” is mentioned in several documents and in multiple different languages, the tool knows that these all refer to the same entity and will incorporate this information into the visualization.

Rosoka relationship extraction includes roughly 500 relationship types out of the box, and can incorporate user defined relationship types as well. Each relationship type can also include subtypes; for example, a Person to Person relationship may have a type of “familial” and a subtype of “parent_of”. Rosoka relationships are outputted as Predicate-Subject-Object statements, or PSOs, which can easily be read into existing triple stores.