Reston, VA – Finch Computing, makers of the innovative Finch for Text® natural language processing software, today announced that the proprietary solution now includes “products” as a supported entity category.
“Beyond entity intelligence about people, places, organizations and more, the addition of product as a supported entity type enables us to deliver an even more customizable – and, as a result – valuable solution to our customers,” Finch Computing Chief Technology Officer Scott Lightner said.
Finch for Text®, which uses natural language processing to extract, enrich and disambiguate entities in text, among other capabilities, delivers real-time, at-scale entity intelligence to customers in the intelligence, national security, financial and commercial arena. By adding “product” to the supported entity types in the product, it can now offer even more dimensionalized insights to customers.
As a jumping off point for the addition of products, while working with a customer supporting the U.S. Air Force, the Finch team trained its proprietary algorithms to recognize more than 12,000+ commercial and military aircraft, including 390 distinct engines and 545 missile systems. The team then developed a product knowledge base that included thousands of features about the entities, such as manufacturers, designers and more. This feature knowledge enabled Finch for Text® to correctly recognize aircraft mentions in text and to distinguish them from any other entity or mention of the same word. As an example, understanding that a reference to “stealth” was in fact a mention of an aircraft and not merely a descriptive word appearing in text.
The model training and tuning necessary to develop this sub-category of product is now being used to add other products to the Finch for Text® entity knowledge base, which includes millions of entities and billions of distinguishing features about those entities.
To learn more about Finch Computing, including about the IP library that enables our unique approach to entity intelligence, please visit www.finchcomputing.com.