At the core of Sinequa's architecture is our patented Context EngineTM. Combining semantics with statistics - in other words, natural language processing and analysis - results in smarter answers, delivered faster. Powerful enterprise search means understanding context - not only the context of the immediate question, but also the context of the person asking it.

Sinequa's users see that they simply get better answers, faster. The enterprise, in turn, sees better productivity and agility. Here's the reason why: Better technology, developed specifically for business needs.

For years, companies have attempted to improve productivity and business value through enterprise search. But the nature of user needs within a company differs from the popular, statistics-based search technologies that work for the Web, limiting their relevance. You don't search for the history of client interactions, recent transactions or contact persons the same way that you search for a supplier's address  or a product datasheet.

Some business queries occur just once, but their business impact - at the very moment an important decision must be made - can be tremendous. Statistics are far too limited in this environment; therefore, performance and results depend on the built-in intelligence of the search engine designed specifically for business.

We offer the first and only enterprise search approach based on a dynamic combination of four types of indexing technology: statistical, structured, linguistic and semantic. The Sinequa difference - the semantic index - is based on a global dictionary in six languages, in which each word is associated to a position in a vector space of 800 dimensions. Moreover, the Sinequa context engine generates additional metadata linking to documents and relative business objects (e.g., a company mentioned in a document is linked to the same company's CRM record). This provides an unparalleled level of intelligence and capacity to correlate information and data.

Here's the technical context behind a delivering better results - and it starts from the instant a user enters a query:

1. Semantic Analysis	Sinequa understands natural language Text conversion to HTML format and UTF-8 encoding Punctuation, marks detection & word separation, cleansing of tags Automatic language detection Lexical analysis: Part of speech tagging, compound word detection Syntactic analysis: Disambiguation, lemmatization of nouns, verbs and adjectives Semantics: Determining the various meanings of each word, then of the whole text Indexing base forms, original forms, proper nouns and the semantics vector
2. Query Processing	Sinequa analyzes context Query cleansing Automatic language detection Lexical analysis: Part of speech tagging, compound word or empty word detection Query Boolean operators (+, - , "…", *, …) Query expansion by fuzzy search, phonetics Query expansion by standard related terms (noun/verb, noun/adjective, acronyms, etc.) Query expansion by additional dictionaries (thesaurus, multilingual lexicon, etc.) Semantics: Determining the various meanings of each word, then of the whole text
3. Knowledge Extraction	Sinequa delivers more relevant results every time Fine tuning of relevance (lexical, weighted, multiple information retrieval techniques, etc.) Display relevant extracts and metadata Dynamic concepts extraction and refinement Common facet navigation based on available metadata (size, format, language, source, category, author, etc.) Standard named entities extraction (people, companies, locations) User-defined named entities extraction Text mining agents: Dynamic relationships and complex pattern extractions Classifiers: Dynamic classification of documents Rules-based taxonomy