The objective of the proposed research is to develop models of human computer interaction for exploratory data analysis. The research will be focused on the information presentation part of the interaction, especially:

1) interactive exploration of behavior data sets about dynamic systems that include spatial and temporal references;

2) automatic generation of natural language presentations from these data sets (e.g. summaries, comparisons) that are coordinated with mixed media (e.g. charts, timelines, maps, animations, etc.); and

3) adaptation of the generation process to the user and the interaction history. For this purpose, problem solving methods from the field of artificial intelligence will be used to support reasoning for abstraction, planning and configuration during the dialog process.

Reusable knowledge sources will also be used in the form of available domain ontologies and knowledge extraction techniques from text and speech corpora in the development of specific components for dialog management and language generation, specifically freely available resources related to behavior data of dynamic systems (e.g., temporal knowledge, spatial knowledge, qualitative physical models, etc.). The research will follow a methodology based on two loops of design-implementationevaluation of experimental prototypes: the first loop will be carried out during the ongoing phase and the second during the return phase. The goals of the research project include the generation of the products during the two-year project. These products include:

a) two corpora for experimentation purposes for two different domains based on real world problems;

b) two software experimental implementations supported by algorithms corresponding to the inference procedures and reusable domain knowledge with its corresponding declarative representation;

c) the results of the evaluation on both domains.