eXenSa will help you make sense of the tons of data your systems collect every single second. We have developped a fantastically powerful method that can make a difference in many different domains. While initially targeting textual data, this method can also be used for the collaborative filtering problems : what gives birth to the right recommendations to your customers, for instance.

Because our engine (ChEn) is fast and scalable, and can produce sense out of data, we can offer a large panel of applications :

e-Commerce recommandation system

For all the e-Commerce sites on the 'net, the Graal is to be able to propose a personal interface to its returning customers, with the right products pointed out. What is even more looked after is a way to quickly guess the goal of the visit of an unknown visitor, in order to present her with the products she is the most likely to look for. By creating a large numbers of profiles, relating product visits with users, with basket composition and with sales, and filtering this data with our engine, we will serve you the very best recommendations. Available in Q3-2011.

More to come :

Content recommandation system

Similarly to the e-Commerce recommandation system, content sites (news, forums, encylopediae) will have the opportunity to present the most adapted content to their users, in order to improve the experience, and make their visits last longer. By knowing who consults/advices what, who appreciates / dislikes what, our engine will be able to quickly draw a profile for new visitors, and to produce a ranked list of the most adapted content for each user and visitor. Available in Q3-2011.

Tailored search engine

Do you think a generic search engine (even a self proclaimed "semantic" search engine) can really serve your needs ? Think again : each community develops its language (ever noticed SMS-language, chat-only acronyms, l33tsp34k, lolcat-speak ?) , each field of interest has a very personal semantic, each site has a different public. We are developping a dynamic search engine that you can control : it can learn from every new document created on your site, in can learn from who reads what, who advices what, and so on. Each query on this search engine will be exactly tailored for your site, for the user and for your documents.Available in Q2-2012.

Our technology is far beyond the classical and pervasive LSI (or its probabilit-based alernative, pLSI) used by the biggest search engines. It's actually better than most advanced neuromimetics models, while being way faster and not limited by the input dimensionality.