Modeling a Probabilistic Ontology for Maritime Domain Awareness

2009

Achieving agile Command and Control in Maritime Operations requires composeability-the ability to construct responses "on the fly" to queries about a given situation, by discovering and drawing upon the appropriate resources from among the vast collection of resources existing on a distributed network. Although net-centric architectures such as FORCEnet provide the necessary connectivity and computational power needed to achieve fast and adaptive decision cycles, the sheer volume of data creates informational and cognitive bottlenecks that hinder agility. To address these limitations, new approaches bridging the gap from data interchange to knowledge interchange are needed, enabling C2 systems to produce a dynamic, comprehensive, and accurate battlespace picture. This is the main focus of PROGNOS, a system for Predictive Naval Situation Awareness currently being developed at George Mason University's C4I Center. PROGNOS will integrate four state-of-the-art enabling technologies into a distributed system architecture that represents domain knowledge as a modular collection of probabilistic ontologies, combine these "knowledge nuggets" dynamically into complex situation models, and apply theoretically sound, computationally efficient hypothesis management and inference to combine evidence and background knowledge to reason about the current situation. PROGNOS will also interoperate with other FORCEnet systems by interacting via semantically enabled services.

2012 Oceans - Yeosu, 2012

The upsurge in piracy and the impact of recent environmental disasters have highlighted the need to improve maritime surveillance. Governmental and private initiatives have developed monitoring systems with improved acquisition and analysis capabilities. These systems rely on one major component, namely the detection of abnormal ship behaviour. This implies a detailed formalisation of expert knowledge. However, the quantity of data, the complexity of situations, the failure to take into account their spatial characteristics and the potential for the same scenario to be interpreted in different ways have proved to be significant problems. We therefore propose a new prototype for the analysis of abnormal ship behaviour. The system is based on a spatial ontology associated with a geographical inference engine. It automatically identifies suspicious vessels and associates them with probable behaviours defined by operational staff.

Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number.

2012 15th International Conference on Information Fusion, 2012

Maritime information systems typically offer system features for data collection, information fusion and the automated construction of a basic recognized maritime picture for human operators. However making sense of maritime situations is still very much an intensive human-based cognitive endeavor. Limited human resources face difficult cognitive challenges in making sense of huge amount of data generated daily by dynamic roundthe-clock maritime shipping and port activities. An effective information exploitation system for maritime surveillance and monitoring is expected to be able to augment human-based surveillance and monitoring operations with machine-based computing capabilities to interpret and reason about massive amount of situational information, as well as transforming this knowledge into actionable decision indicators for the human decision makers. The Defence Science & Technology Agency (DSTA) has developed the PACKED model to guide the exploration and development of computing technologies that support human decision making and other cognitive challenges in information rich, complex, and dynamic operational environments. An integrated maritime reasoning and monitoring system illustrates how computing technologies are applied in the areas of Attention, Knowledge, and Comprehension to support situation analysis and monitoring processes for maritime surveillance. The maritime reasoning and monitoring system consist of Bayesian reasoning, entity network analysis, and movement pattern analysis, to integrate, analyze, interpret and reason about maritime data. The Graylist Network Analysis (GNA) engine probes networks of entity-relationships to uncover "graylisted" entities that are closely associated with blacklisted entities. The Movement Pattern Analysis (MPA) engine uncovers common patterns in vessel movements and monitors real-time for maritime traffic data for anomalous vessel movements. The Bayesian network inference engine combines situation information from data sources and analysis results produced by analytical engines such as the GNA and MPA, with Bayesian knowledge models in order to infer probabilities of occurrences of maritime scenarios.

2013

Proceedings of: NATO Advanced Study Institute (ASI) on Prediction and Recognition of Piracy Efforts Using Collaborative Human-Centric Information Systems, Salamanca, 19-30 September, 2011Ontology-based representations are gaining momentum among other alternatives to implement the knowledge model of high-level fusion applications. In this paper, we provide an introduction to the theoretical foundations of ontology-based knowledge representation and reasoning, with a particular focus on the issues that appear in maritime security –where heterogeneous regulations, information sources, users, and systems are involved. We also present some current approaches and existing technologies for high-level fusion based on ontological representations. Unfortunately, current tools for the practical implementation of ontology-based systems are not fully standardized, or even prepared to work together in medium-scale systems. Accordingly, we discuss different alternatives to face problems such as sp...

2012

The paper presents 3 key technologies used in an integrated maritime reasoning and monitoring system [1]. These technologies fuse maritime data for decision support in the martime security domain. They support monitoring of scenarios of interest such as terrorist attacks or piracy incidents, and alert operators when such scenarios are assessed likely to happen. These 3 technologies are: a. Dynamic Bayesian Reasoning & Advanced Intelligent Networks (D'Brain), a reasoning system based on Bayesian networks. Causal and probabilistic information about scenarios of interest are structured as Bayesian networks, in which variables are set according to the interpretation of incoming situation information. The system thus provides a systematic approach to modelling and monitoring for scenarios of interest in the maritime situation. b. Graylist Ranking, a ranking technique to order entities based on their structural relationship with black-listed entities. The ranked list of entities could then be used to prioritize further investigation efforts into higher ranking entities which are considered more suspicious. c. Trajectory Analysis using a nonparametric Bayesian model, Dual Hierarchical Dirichlet Processes (Dual-HDP) [2]. It models the activity patterns (norm) of entities in motion. With the norm established, anomaly detection could be performed by evaluating how much each track deviates from the norm.

2010

Systems are increasingly required to fuse data from geographically dispersed, heterogeneous information sources to produce up-to-date, mission-relevant results. These products focus not only on traditional military forces and systems, but to an increasing degree also on non-traditional combatants and their social networks. Successful multi-INT fusion requires that the constituent systems interoperate not just at the level of syntax and formats, but also at the level of semantics. Ontologies are vital enablers for semantic interoperability. Because uncertainty is a fundamental aspect of multi-INT fusion, lack of support for uncertainty is a major limitation of current-generation ontology formalisms. Probabilistic OWL (PR-OWL) extends the OWL Web Ontology Language to enable the construction of probabilistic ontologies. Ontologies constructed in PROWL can represent complex patterns of evidential relationships among uncertain hypotheses. Recently, a system for specifying and reasoning with PROWL ontologies has been released in alpha version. This paper describes the PROWL ontology language, the probabilistic logic on which it is based, and the reasoning system implementation. A hypothetical case study in the counterterrorism domain is illustrates the capabilities of PROWL .

Information Fusion, 2013

The concepts of event and anomaly are important building blocks for developing a situational picture of the observed environment. We here relate these concepts to the JDL fusion model and demonstrate the power of Markov Logic Networks (MLNs) for encoding uncertain knowledge and compute inferences according to observed evidence. MLNs combine the expressive power of first-order logic and the probabilistic uncertainty management of Markov networks. Within this framework, different types of knowledge (e.g. a priori, contextual) with associated uncertainty can be fused together for situation assessment by expressing unobservable complex events as a logical combination of simpler evidences. We also develop a mechanism to evaluate the level of completion of complex events and show how, along with event probability, it could provide additional useful information to the operator. Examples are demonstrated on two maritime scenarios of rules for event and anomaly detection.