Automated detection and exploration of known significant temporal patterns, and discovery and exploration of new meaningful temporal patterns in time-oriented quantitative and qualitative raw data from multiple sources
Researcher: Yuval Shahar
Department: Information Systems Engineering
Faculty: Engineering
E-mail: yshahar@bgu.ac.il , ysmagic@yahoo.com
Interpretation and exploration of longitudinal data arriving from multiple sources is a major part of decision making in most tasks, in particular monitoring and planning. Such tasks are crucial, for example, in intelligence gathering, in particular for integration of data arriving from multiple channels and for making decisions based on the interpretation and exploration of heterogeneous data.
We are developing the Distributed Intelligent automated Monitoring and Exploration [DIME] architecture. DIME is a distributed knowledge-based computational architecture specific to the tasks of monitoring, interpretation, and interactive exploration of large numbers of distributed time-oriented data, using an innovative, effective visualization interface. The DIME framework is unique in enabling users to explore both raw data and multiple levels of concepts abstracted from these data, by moving through a network of domain-specific concepts and relations. The abstraction, visualization, and exploration processes will access a set of distributed databases and a set of domain-specific knowledge bases that will be maintained by domain experts using a task-specific knowledge-acquisition tool. The architecture supports runtime data and knowledge integration of multiple remote heterogeneous time-oriented databases. A preliminary architecture has been implemented and has been evaluated in the medical domain of monitoring chronic patients. The architecture is based on our extensive previous theoretical and practical research in the areas of knowledge-based temporal reasoning and intelligent user interfaces. Additional computational and interface enhancements have been added over the past several years, and the resultant prototype architecture has been installed in an intelligence-gathering site. Currently, we aim to capitalize on the theoretical and practical temporal-reasoning, temporal data-mining, and visual analytical work that we have been doing so far, so as to build a computational box that integrates both the data driven (continuous monitoring) and goal-directed (user-initiated query) computational modes, and thus provide an answer to both of these intelligence and security needs. We also aim to provide continuous, data-driven, pattern-matching capabilities, and visualization and exploration tools that enable users to visually query and explore the resultant abstractions and patterns for a whole population of monitored subjects.
Cultural Intelligence and linguistics
Researcher: Yair Neuman
Department: Education
Faculty: Humanities and Social Sciences
Email: yneuman@bgu.ac.il
Prof. Yair Neuman is an interdisciplinary researcher who merges ideas from psychology, linguistics and natural language processing for developing novel technologies that identify meaning emerging from textual data. Currently he is a PI on a project sponsored by IARPA. He also developed a new technology for CULINT (Cultural Intelligence) as well as a methodology for identifying themes in textual data.
