Our main interest is the development and application of methods for mining data from different domains, most prominently biological and social data. This is a wide-spectrum project: it cannot be characterized by involving a single application domain, neither a single type of data, neither a single type of mining/learning method. Most modern tasks in data mining, even for a specific domain, require using a large repertoire of methods and dealing with a variety of forms of information.

This project can largely be viewed as a continuation of a series of projects similar in nature and goals to this one by two of current participant sites (Zaragoza, Barcelona) and including also some participants in the new coming site (Cantabria). The scope of these projects started being mostly theoretical and has acquired more applied aspects over time. The current project is a further step in this direction: while remaining firmly grounded in principled, mathematically flavored research, it aims at solving real-world problems.

Specifically, the BASMATI project aims at

1. developing new machine learning and data mining algorithms of practical applicability
2. demonstrating such applicability through implementations tested on practical tasks, by investing as much adaption effort to the target fields as necessary; while at the same time
3. creating theoretical tools that provide formal guarantees of the performances to be expected in the application of these algorithms, both in terms of efficiency and of predictive or descriptive power. Our main target application fields will be biological and social data.