|
contents | software | |||||||
| Basis Technology's Rosette Linguistics Platform 5.0 Basis Technology has announced the availability of release 5.0 of its Rosette Linguistics Platform (RLP). This release includes a wide range of new features and performance enhancements to Rosette's core technology, designed to provide a fast, accurate, flexible, and easy-to-integrate solution for tackling the most complex linguistic challenges. New features and improvements include: - Rosette Entity Extractor (REX) - which identifies names, places, dates and other entities in unstructured text - has added Traditional Chinese to its list of ten supported languages, and improved its overall speed by an average of 35 percent. Specific language accuracy was also improved, highlighted by Chinese and Arabic, which increased by 27 percent and 12 percent, respectively. - Rosette Language Identifier (RLI) - which identifies the language and encoding of a document - has improved accuracy on short text segments which help with the analysis of smaller documents, and has added Pashto, Somali, and Urdu detection. - The addition of a Japanese Orthographic Analyzer (JOA) which allows search engines and text mining applications to find occurrences of words that have multiple spelling variations but shared meaning, a common and frequent instance in the Japanese language. JOA enhances RLP's Base Linguistics component which performs critical functions such as word segmentation, decompounding, and part-of-speech analysis. - Newly updated and greatly expanded Chinese and Japanese dictionaries. The Rosette Linguistics Platform is a suite of interoperable components for applications that analyze and process all the world's languages. RLP enables information retrieval applications and search engines to understand queries and search web pages in foreign languages by providing a linguistic analysis of each word. Each product incorporates an extensive knowledge base of the nuances of each language, enabling applications to deliver reliable, accurate search results in many languages. write your comments about the article :: © 2006 Computing News :: home page |