Top Ten Trends in High-Level Information Fusion Erik Blasch AFRL/RIEA Rome, NY, 13441 erik.blasch@rl.af.mil Pierre Valin Anne-Laure Jousselme Defence R&D Canada-Valcartier pierre.valin@drdc-rddc.gc.ca Dale Lambert Defence Science & Tech. Org. Edinburgh, SA, Australia dale.lambert@dsto.defence.gov.au Éloi Bossé Université Laval Québec, Canada ebosse@gel.ulaval.ca Abstract – High-Level Information Fusion (HLIF) is a relatively new exploration of methods in the last decade. The discussion will address the issues between low-level (signal processing and object state estimation and characterization) and HLIF (control, situational understanding, and relationships to the environment). From a series of efforts in identifying the main research focuses for the next decade, we have identified the main issues from fusion conference papers and panel discussions, towards a comprehensive analysis. With the advent of the HLIF grand challenges, many of the issues were analyzed over the last decade. In this paper, we highlight the main themes and a discussion of the attributes of the top ten issues. Since IF is to reduce uncertainty, a focus of this paper for the Evaluation of Techniques for Uncertainty Representation (ETUR) working group is to posit the issues of uncertainty for HLIF. Specific trends include data/knowledge representations, situation/threat/impact assessment, systems design, evaluation, and information management. The paper concludes with a topic of brief analysis of an uncertainty ontology for the ETURWG. Keywords: Fusion, Situational/Impact Assessment, Resource/Sensor Management, User Refinement 1 Introduction High-level Information Fusion (HLIF) has been of considerable interest to the fusion community ever since the development of the fusion process models. The low- level versus high-level distinction was made evident in the seminal text on the subject by Waltz and Llinas, Multisensor Data Fusion. [1] While many discussions in HLIF have been coordinated in the past decade at the fusion conferences, including other panel discussions, there is a need to gather contemporary insights into the ongoing challenges. Recent HLIF texts include: Mathematical Techniques in Multisensor Data Fusion [2], Concepts, Models, and Tools for Information Fusion [3], High-Level Fusion [4], Handbook of Multisensor Data Fusion, [5-6], Brain-mind Machinery [7], and High-Level Information Fusion Management and Systems Design [8]. Many panel sessions at the International Conference on Information Fusion (ICIF) have focused on HLIF of which we sought to canvass [9] and organize the discussions on HLIF. HLIF typically includes situation/threat/impact (SA/TA/IA) assessment and resource management. For the general problem areas, we further analyzed the results from which ten themes emerge as categorized below: Area A. Data/Knowledge Representation Reference Model Taxonomy of notations, symbols, and meanings Area B. SA/TA/IA Assessment Semantics/ontologies Social/Behavioral/Cultural Models Area C. Systems Design User/agent coordination Display (interactive) Area D. Evaluation Common scenario, Performance comparison Metrics / Uncertainty analysis Area E. Information Management Resource planning and information analysis Joint theory of methods integration As related to the Evaluation of Techniques for Uncertainty Representation (ETUR) Working Group, we note that uncertainty analysis and ontologies has not been given enough consistent attention. The main focus on panel discussions have been on what is HLIF, how to measure HLIF, and the coordination of the machine and the user. Figure 1 presents an overview of uncertainty in a system as related to the above areas where the situation, depicted at the bottom, is affected by the assessment evaluation, representation, machine (information management), as well as the human (user coordination). Figure 1. Situation Uncertainty To derive the top trends in HLIF, numerous efforts were conducted to bring together the new ideas in HLIF over the last decade which featured problem descriptions, challenges, and issue discussions. 1.1 Lambert’s Grand Challenges Dale Lambert [10] posed some HLIF grand challenges for the Information community in 2003 to include: E. P. Blasch, P. Valin, A-L. Jousselme, D. A. Lambert, and Éloi Bossé, “Top Ten Trends in High-Level Information Fusion,” International Conf. on Information Fusion, 2012.