An Adaptable Platform for Text Analysis and Discovery
Mission-critical AI for human language, deployable in any environment
Who Uses Rosette
Text analytics by Rosette® is making it possible to find signals from social media and big data, whether the purpose is performing due diligence, learning what makes customers happy and unhappy, or forecasting imminent events. Rosette’s intelligent name matching technology is enabling efficient and accurate watchlist screening, and patient/customer records lookup.
What Rosette Does
Our broad language coverage is also deep. We don’t release a new language until its out-of-the-box quality and performance meet our tough standards. Most importantly, customers are empowered to configure and customize our software to adapt it to their data and use case for the best possible results.
How Rosette Does It
Our natural language processing technology uses the method best suited to each task, whether machine learning, rules, or dictionaries. We continually evaluate, develop, and refine our technologies to meet tomorrow’s needs.
Basis Technology has been enabling customers to verify identities, understand customers, anticipate world events, and uncover crime. Since 1998, Rosette has been adopted by over 200 major firms, including Amazon.com, Société Générale, Microsoft Bing, Yahoo!, Oracle, Hitachi, Fujitsu, and government agencies around the world. Rosette is also fueling the current generation of innovation including Airbnb, Luminoso, Medallia, Pinterest, Recorded Future, and Yelp.
Fundamental morphological analysis in 30+ languages to prepare your data for analysis: tokenization, lemmatization, part-of-speech tagging, noun decompounding, etc.
Classifies documents by topic or taxonomy. Easily trained to support your own categories.
Converts Arabic written in Latin script (Sub7anallah) back to Modern Standard Arabic (سبحان الله) for automated text analysis.
Finds the people, organizations, locations, and other significant entities mentioned in your text for data triage, metadata creation, and more.
Distinguishes between similarly named entities by linking each one to a knowledge base (yours or Wikidata) of people, organizations, and locations (and 700+ subtypes).
Tags the language (from 55+) of each document or multiple languages in one document. Detects 25+ languages given as little as 1-3 words.
Matches names of people, organizations, and locations across languages and scripts, misspellings, nicknames, initials, titles, misordered name components, etc.
Translates names consistently and swiftly from one script to another, including English, Arabic, Chinese, Japanese, Korean, Persian, and Russian.
Extracts targeted personal and organizational relationships between entities. Customizable to find entities connected by other types of relationships.
Makes cross-language search and duplicate document detection a reality with text embeddings that find words with similar meanings across languages.
Discovers positive, negative, or neutral sentiment in a document, or towards a person, place, or thing.
Identifies keyphrases that summarize a document, and find concepts even when they are not specifically named.