Temporal/ Multi-dimensional Reasoning with Imprecision and Uncertainty
Research Project

Debasis Mitra, Principal Investigator


Host Institution: Florida Institute of Technology
Funded By National Science Foundation (IIS-0296042)

Abstract


The goal of this project is to develop algorithms for the purpose of reasoning with time and other higher dimensional space. Reasoning with time and three-dimensional space is ubiquitous in our daily life as well as in computer systems. For example, a detective would often check the temporal consistency of alibi of some suspects. A system for packing boxes in a container would reason with space. For about last two decades some researchers within the artificial intelligence community are engaged in studying this area. The PI has been involved in empirically analyzing some of the existing algorithms and developing new efficient algorithms for temporal reasoning. A particular area of interest in this project is on studying temporal reasoning under uncertainty. If the input information is not completely known and is rigged with uncertainty, then how does it affect the reasoning process? How to guess possible causes for events from the available temporal information? These are the types of questions, which are being investigated in this work. Viewing temporal (or higher dimensional) information as constraints, an algorithmic approach for constraint propagation is being employed in the research. A deeper understanding and efficient mechanisms for dimension related reasoning would emerge from this project.

AN INTRODUCTORY PRESENTATION IS HERE.

Theoretical Challenges Addressed

Investigating the foundations of differnent space and time related constraint algebras.

Investigating points in 2D and nD spaces.

Investigating partially-ordered time domains.

Investigating meaningful sub-algebras (e.g., causal) of the interval algebra

Generating all consistent scenarios

Investigating uncertainty-based dimension-related reasoning


Targeted Application Areas

(Moving targets!!)

Developing an integrated system for different ontologies of time and space for both qualitative and quantitative reasoniong.

Visualization of historocal information.

Multi-dimensional information modeling for "intelligence" related data analyses.

Geometrical modeling..


Final Report submitted to NSF.


Useful Links

Publications

E-mail: dmitra@cs.fit.edu
Disclaimer:
National Science Foundation and Florida Institute of Technology are not responsible for any information contained here, and I am not responsible for any information one could access from the links on this web page.