Parity's Profiler System assembles and analyzes
detailed profiles of key entities such as people, institutions, and products,
from disparate unstructured documents and semi-structured data sources.
Profile assembly is driven by client-specific data needs specifying
the key entities to be profiled, important attributes to collect, and
types of relationships between the entities to infer. The key entities
are extracted and assembled into distinct profiles using advanced machine
learning heuristics. This includes normalization of spelling variations
together with disambiguation of similarly-named entities (e.g. two people
with the same name).
The Profiler System discovers relationships between entities, such as
affiliations, citations between documents, and business relationships.
The result is an accurate and comprehensive profile for each distinct
entity with detailed relationships to other entities, and links back to
the source databases and documents.
This enables automated analysis of the profiles and detailed matching
to desired criteria, going beyond keyword-spotting techniques of standard
- Automated solution: Profile assembly is a fully
- High accuracy: Assembled profiles have very high
precision in matching and disambiguating between same name entities.
- Client specific models: Profiles are constructed
to fill out client's knowledge needs. The data models capture the key
entities of importance, which may include documents, people, companies
and other institutions, products, and other application-specific entities.
- Multiple data sources: The Profiler System builds
profiles from disparate unstructured text documents, semi-structured
data with textual fields, and OCR documents. Profile data can be extracted
from databases and documents from inside the organization, but can also
be combined with profiles built from external sources including web-based
data, news articles, and outside reports.
- Network of links and relationships: The Profiler
System derives links and relationships between profiled entities. The
links can be both explicit, where there is an explicit relationship
between the two profiles or implicit, where the profile relationship
is derived from domain specific heuristics.
- Advanced search and navigation: The Profiling System
provides APIs to enable domain-specific retrieval of profiles and navigation
of the knowledge network. This includes specialized search engines to
index and retrieve profile entities as well as customizable navigation
of profile relationships.