We are pleased to present the proceedings of the 12th International Conference on Discovery Science (DS 2009), held in Porto, Portugal, October 3-5, 2009. DS 2009 was collocated with ALT 2009, the 20th International Conference on AlgorithmicLearningTheory,continuingthesuccessfulDSconferenceseries. DS 2009 provided an open forum for intensive discussions and the exchange of new ideas among researchers working in the area of discovery science. The scope of the conference included the development and analysis of methods for automatic scienti?c knowledge discovery, machine learning, intelligent data analysis, and theory of learning, as well as their applications. We were honored to have a very strong program. Acceptance for the conference proceedings was very compe- tive. There were 92 papers submitted, with the authors coming from roughly 20 di?erent countries. All paperswere reviewedby three senior researchersfollowed by an extensive discussion. The program committee decided to accept 23 long papers (an acceptance rate of 25%) and 12 regular papers. The overall acc- tance rate was 38%. The contributed papers cover a wide range of topics, from discovery in general to data mining in particular. In addition to the technicalpapers, we weredelighted to have?ve prestigious invited speakers and two tutorials. Fernando Pereira, University of Penns- vania, USA, presented new fundamental questions that should be investigated in natural language processing in web mining. Hector Ge?ner, from Pompeu Fabra University, Spain, discussed learning methods for solving complete pl- ning domains.