Reston, VA – Finch Computing, developers of the innovative, real-time natural language processing solution Finch for Text®, today announced that it has added relationship extraction and co-references to the product. Relationship extraction gives users an ability to decipher relationships between entities, and co-reference enables words like “her” or “him” or “the leader” appearing in text to be resolved to a specific, named entity.
“These two new features will give Finch for Text® users even more intelligence about the entities they’re interested in while also making the product even more accurate and valuable than it already is,” Finch Computing Chief Technology Officer Scott Lightner said.
Finch for Text® can now find important relationships between entities such as: Acquired-by, Co-Investor- with, Competitor-with, Customer-of, Director-of, Educated-at, Employer-of, Founder-of, Invested-in, Organization-Location, Owner-of, Partner-of, Person-Location, Relative-of, and Subsidiary-of. For companies and people in particular, understanding these connections helps users perform faster, richer and deeper analysis.
Entity co-reference refers to the ability to resolve otherwise obscure references to an entity – like her, him, the company, the product – to a disambiguated entity. The value of this capability is that it helps users understand all mentions of an entity even if that mention isn’t by name. It improves salience scores because the product can better gauge how much an article is about a given entity. It also improves sentiment scores with more mentions to analyze, and the same is true for relationship extraction – there are more relationships discovered because there are more mentions linked to an entity.
“We are thrilled about this evolution of Finch for Text®’s capabilities and we look forward to continuing to delight our customers with entity intelligence and accuracy that is unrivaled in other products,” Lightner continued.
For more information about Finch Analyst and to request a free demo account, please visit text.finchcomputing.com.