SESAAME (Secuencias Simbólicas: Análisis, Aprendizaje, Minería y Evolución)
SESAAME is a three-year long (2009-2011) project funded by the Spanish Ministry for Science and Innovation
Its name is an acronym for Spanish "Symbolic Sequences: Analysis, Learning, Mining, and Evolution".
SESAAME is a coordinated project among three sites:
- Universitat Politècnica de Catalunya (Site and projecte coordinator: Ricard Gavaldà)
- Universidad de Zaragoza (Site coordinador: Elvira Mayordomo), with affiliated researchers at Universidad de Valladolid and Universidad Carlos III in Madrid.
- Universidad de Málaga (Site coordinator: Rafael Morales), with affiliated researchers at University of Porto
A full description of the project can be found in the technical proposal. The research goals of the project are summarized as:
- Develop methods for model comparison, evaluation, and selection that can be used for in algorithms for explaining, classifying, and predicting massive datasets. Apply them specifically to multiclassfier (ensemble) methods.
- Study the fundamentals and design methods for analysis, learning, and mining data: Develop parallel systems for time series processing. Develop and test a new methodology for recommender systems using tools from FCA.
- Develop further methods for analysis and mining in sequential and structured data: Develop more efficient algorithms for inferring Markov models and probabilistic automata from data. Algorithms for minng frequent patterns and combinatorial structures from data. Compression, online prediction, and dimension of geometrical data and XML.
- Develop techniques and theoretical foundations of learning in Data Streams. Study the quality of generated models by learning algorithms and develop algorithms that can learn in time-evolving. Formalize data streams as a computation model.
- Apply the methods in real contexts. Development of a blog analyzer; a tool for assisting in email classification, grouping, classifying, and prediction of genomic sequences; and analysis of web data and applications of data mining to autonomic computing.
- Study human and animal language as sequential processing. Study and extension of grammar models and language games.
- Study of evolutionary mechanisms in systems of agents involving sequences, in particular linguistic communities and biological evolution, especially the concept of modularity.