Banter Inc., which specializes in natural language processing, this week released an updated version of its Relationship Modeling Engine (RME).
The San Francisco-based company says RME v 5.0 can be easily embedded in any software application that handles unstructured information and free text. The software fully supports UNICODE and ASCII for single byte applications.
This is necessary because everyday language analysis is different, and more complex, than the analysis of well-written and relatively structured documents. The software can analyze and understand the content of documents and messages written in everyday language, including the naturally occurring variations in expressions, spelling and grammar
Typical uses for RME range from e-mail analysis, chat support, online self-service and natural language information retrieval, to document classification and tagging. Companies like Nintendo, Wells Fargo and VeriSign are currently using the technology.
"With a broader and deeper scope, RME version 5.0 is able to deliver a higher level of accuracy in analyzing everyday language for the ever-expanding volume of human communication and requires minimal administrative effort and IT maintenance," says Banter founder and chief technology officer Yoram Nelken.
Nelken says RME 5.0 also includes clustering, (the grouping of uncategorized texts into "natural" categories), hierarchy sub-tree matching; multi-taxonomy support and concept highlighting and extraction, (highlighting the segments in a text that caused a specific category to be selected or not to be selected).