Smart Information Flow Technologies - plan recognition

LTML - A Language for Representation Semantic Web Service Workflow Procedures

admin — Sun, 02 Aug 2015 19:36:18 +0000

Keywords: insider threat detection, network security, plan recognition

Abstract: The Learnable Task Modeling Language (LTML) was developed by combining features of OWL, OWL-S, and PDDL, using a more compact andbreadable syntax than OWL/RDF to create human readable representations of webvservice procedures and hierarchical task models. Our goal was in part to develop a more robust and developer-friendly language based on the principles and design that led to OWL-S and demonstrate that such a language also provided the basis for developing tools that could learn web service procedures by demonstration. LTML’s initial and driving use is as an interlingua for the learning and procedure execution components of POIROT, a system that learns web service workflow procedures from ‘observations’ of one or a small number of semantic web service traces. The LTML language uses an s-expression based syntax for improved readability but has parsers and generators that translate the surface forms into RDF for storage in a SESAME triple store implementing POIROT’s internal blackboard. All language elements are grounded in a set of OWL ontologies. The language encompasses and extends coverage of the OWL-S process and grounding models, and introduces elements to support sets of hierarchical task methods indexed by goals, semantic execution traces, and internal tasks and learning goals. This short paper gives an overview of LTML and describes the areas where LTML diverges from or extends OWL-S and PDDL.

Citation:

Burstein, M., Goldman, R., McDermott, D., McDonald, D., Beal, J., & Maraist, J. (2009, October 26). LTML — A Language for Representing Semantic Web Service Workﬂow Procedures. Workshop on Semantics for the Rest of Us: Variants of Semantic Web Languages, Workshop conducted in conjunction with 8th International Semantic Web Conference (ISWC 2009), Washington, DC.

Date Published:

Thursday, October 1, 2009

Publication File:

LTML.pdf

Workflow: Editor Role:

Request Review

Workflow: Publisher Role:

Publish

Prevent Workflow Messages:

Trapping Malicious Insiders in the SPDR Web

admin — Sun, 02 Aug 2015 17:37:57 +0000

Keywords: insider threat detection, network security, plan recognition.

Absract: The insider threat has assumed increasing importance as our dependence on critical cyber information infrastructure has increased. In this paper we describe an approach for thwarting and attributing insider attacks. The Sense, Prepare, Detect, and React (SPDR) approach utilizes both a highly intelligent software reasoning system to anticipate, recognize, respond to, and attribute attacks as well as a widely distributed set of hardware-based sensor-effectors to provide alerts used by the reasoning system and to implement responses as directed by it. Using hardware sensor-effectors greatly reduces the risk that a savvy malicious insider can bypass or cripple the system's monitoring and control capabilities. In this paper we describe the prototype SPDR system and the results of its successful evaluation by an independent, DARPA-sponsored Red Team. We conclude with thoughts on possible SPDR enhancements and further research.

Citation:

Haigh, J. T., Harp, S. A., O'Brien, R. C., Payne, C. N., Gohde, J., & Maraist, J. (2009, January 5-8). Trapping Malicious Insiders in the SPDR Web. Proceedings of the 42nd Hawaii International Conference on System Sciences, Waikoloa, Big Island, HI.

Date Published:

Monday, January 5, 2009

Publication File:

2009-hics-spdr.pdf

Workflow: Editor Role:

Request Review

Workflow: Publisher Role:

Publish

Prevent Workflow Messages:

User Acceptance and Plan Recognition: Why Even Perfect Intent Inferencing Might Not be Good Enough

admin — Tue, 28 Jul 2015 20:05:55 +0000

Keywords: plan recognition, adaptive systems, unpredictability, RPA

Abstract: Plan recognition is discussed in the framework of adaptive automation. A case is made for applying the lessons of human-human interactions, such as those found in Cockpit Resource Management, to avoid producing a correct, yet not fully trusted automation partner. Etiquette ‘rules’ for automation are proposed as a mechanism to achieve this.

Citation:

Funk, H., & Miller, C. (2001, November 2-4). User Acceptance and Plan Recognition: Why Even Perfect Intent Inferencing Might Not be Good Enough. Paper presented at the AAAI Fall Symposium on Intent Inference for Collaborative Tasks, North Falmouth, MA.

Date Published:

Friday, November 2, 2001

Publication File:

AAAI-FS-01.pdf

Workflow: Editor Role:

Request Review

Workflow: Publisher Role:

Publish

Prevent Workflow Messages: