Title: IFPDM: a framework to integrate information fusion and probabilistic decision making

Authors: Tamer M. Abo Neama; Ismail A. Ismail; Tarek S. Sobh; Mohamed Zaki

Addresses: The Commandant of the Military Technical College, Kobry Elkobbah, Cairo, Egypt ' Arab Academy for Science, Technology and Maritime Transport, Almushir Ahmad Ismail St., Masakin Sheraton Cairo, Egypt ' Information Systems Department, Egyptian Armed Forces, Omar ebn Elkhattab St., Stage 1, 110 Zahraa Nasr City, Egypt ' Al Azhar University, 1 El Mokhayam Eldaaem St., Nasr City, Cairo, Egypt

Abstract: It is not an easy task to obtain decisions from raw data; therefore, we designed an information fusion and probabilistic decision making framework (IFPDM). It comprises two phases and each phase consists of several layers. The first phase goes up from raw data to belief computation. In this phase, the bottom layer consists of multiple sources of textual data. Then a multi-agent system (MAS) platform is utilised to afford five agents. These agents are organised in a common platform to provide communication, data integration, parser, semantic network and belief computation agents, respectively. The second phase of IFPDM consists of four layers. These layers comprise beliefs combination, pignistic probability computation, probabilistic reasoning and decision ranking, respectively. A case study in which a divide and conquer approach is investigated to prove the concept of IFPDM. It is planned to extend IFPDM to deal with ontology, qualitative beliefs and dynamic Bayesian networks.

Keywords: information fusion; probabilistic decision making; DSmT; Bayesian networks; pignistic probability; multi-agent systems; MAS; agent-based systems; probabilistic reasoning; decision ranking.

DOI: 10.1504/IJIDS.2014.064443

International Journal of Information and Decision Sciences, 2014 Vol.6 No.3, pp.226 - 247

Published online: 10 Sep 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article