The following are sorted by reverse chronological order of publication.
67 publications found:
%0 Generic %A Lowrance, J., Harrison, I., Rodriguez, A., Yeh, E., Boyce, T., Murdock, J., Thomere, J., and Murray, K. %E Okada, A., Buckingham Shum, S. and Sherborne, T. %T Template-Based Structured Argumentation %B Knowledge Cartography: Software Tools and Mapping Techniques %@ 978-1-84800-148-0 %I Springer %S Advanced Information and Knowledge Processing %D 2008 %K Structured Argumentation, SEAS, Evidential Reasoning, Collaborative Analysis
%0 Book Section %A Lowrance, J. and Garvey, G. and Strat, T. %E Yager, R. and Liping, L. (Dempste, A. and Shafer, G., Advisory Editors) %T A Framework for Evidential-Reasoning Systems %B Classic Works on the Dempster-Shafer Theory of Belief Functions %@ 978-3-540-25381-5 %I Springer-Verlag %S Studies in Fuzziness and Soft Computing %P 419-434 %V 219 %D 2008 %K Evidential Reasoning, Dempster-Shafer, Gister %X Evidential reasoning is a body of techniques that supports automated reasoning from evidence. It is based upon the Dempster-Shafer theory of belief functions. Both the formal basis and a framework for the implementation of automated reasoning systems based upon these techniques are presented. The formal and practical approaches are divided into four parts (1) specifying a set of distinct propositional spaces, each of which delimits a set of possible world situations (2) specifying the interrelationships among these propositional spaces (3) representing bodies of evidence as belief distributions over these propositional spaces and (4) establishing paths for the bodies of evidence to move through these propositional spaces by means of evidential operations, eventually converging on spaces where the target questions can be answered.
%0 Journal Article %A John D. Lowrance %T Graphical Manipulation of Evidence in Structured Arguments %B Oxford Journal of Law, Probability and Risk %P 225-240 %V 6 %D 2007 %K structured arguments; evidential reasoning; collective reasoning; collaborative software tool; graphical representations; SEAS %X A semiautomated approach to evidential reasoning uses template-based structured argumentation. Graphical depictions convey lines of reasoning, from evidence through to conclusions. %O
doi: 10.1093/lpr/mgm011
Presented at the workshop on "Graphic and Visual Representations of Evidence and Inference in Legal Settings" at Cordozo School of Law, New York City, Jan. 28-29, 2007. %U http://www.ai.sri.com/pubs/files/1549.pdf
%0 Conference Proceedings %A Yeung, D. and Lowrance, J. %T Computer-Mediated Collaborative Reasoning and Intelligence Analysis %B Intelligence and Security Informatics, Proceedings of ISI-2006 %I Springer-Verlag %D 2006 %K Bias, Intelligence Analysis, Computer Mediated Communications, Collaborative Reasoning, SEAS, Angler %X Problems of bias in intelligence analysis may be reduced by the use of web-based cognitive aids. We introduce a framework spanning the entire collaborative thought process using the Angler and SEAS (Structured Evidential Argumentation System) applications. Angler encourages creative brainstorming while SEAS demands analytical reasoning. The dual nature of this approach suggests substantial benefits from using computer-mediated collaborative and structured reasoning tools for intelligence analysis and policymaking. Computer-mediated communication (CMC) may be impacted by many factors, including group dynamics and cultural and individual differences between participants. Based on empirical research, potential enhancements to Angler and SEAS are outlined, along with experiments to evaluate their worth. The proposed methodology may also be applied to assess the value of the suggested features to other such CMC tools.
%0 Conference Proceedings %A Wolverton, M. and Harrison, I. and Lowrance, J. and Rodriguez, A. and Thomere, J. %T Advanced Patterns and Matches in Link Analysis %B Intelligence and Security Informatics, Proceedings of ISI-2006 %I Springer-Verlag %D 2006 %X The Link Analysis Workbench (LAW) is a tool for detecting and monitoring situations of interest using inexact matching of graphical patterns. Here we describe some recent advances to LAW: incorporating hierarchy, cardinality, disjunction, and constraints in the pattern language and similarity metric, and a flexible, user-friendly interface for displaying matching data. These capabilities support analysts in rapidly exploring and understanding large, incomplete relational data sets.
%0 Conference Proceedings %A Wolverton, M. and Harrison, I. and Lowrance, J. and Rodriguez, A. and Thomere, J. %T Software Supported Pattern Development in Intelligence Analysis %B Proceedings of the IEEE International Conference on Computational Intelligence for Homeland Security and Personal Safety (CIHSPS ’06) %D 2006 %X Intelligence professionals work with incomplete and noisy data. Their information needs are often hard to express, and almost impossible to get right the first time. This paper describes the GEM pattern language for encoding analysts’ information needs in graphical patterns, and its use in the Link Analysis Workbench (LAW) system to find inexact matches to those patterns in large relational data sets. The LAW user typically interacts with the system through a cycle in which the user (1) creates an initial GEM pattern corresponding to his information need, (2) uses the LAW matcher to retrieve a collection of matching episodes in the data, (3) revises the pattern based on the shortcomings of the matches, and (4) repeats the process until the revised pattern is returning the right data. The pattern language and the system are designed to facilitate the user in quickly traversing this cycle. %U http://www.ai.sri.com/pubs/files/1147.pdf
%0 Conference Proceedings %A Rodriguez, A. and Boyce, T. and Lowrance, J. and Yeh, E. %T Angler: Collaboratively Expanding Your Cognitive Horizon %B International Conference on Intelligence Analysis Proceedings %I MITRE %D 2005
%0 Conference Proceedings %A Murray, K. and Lowrance, J. and Appelt, D., and Rodriguez, A. %T Fostering Collaboration with a Semantic Index over Textual Contributions %B AI Technologies for Homeland Security, Papers from the 2005 AAAI Spring Symposium %I AAAI Press %S Papers from the 2005 AAAI Spring Symposium %P 99-106 %D 2005 %X Collaboration is at the heart of many activities required for effective homeland security, from intelligence analysis to policy formation. We are exploring new approaches to facilitating effective collaboration that remove or reduce common barriers and that exploit opportunities to encourage more effective collaboration, including transcending the cognitive biases of the participants. In order to evaluate our approaches we are developing Angler, a web-services tool that supports collaboration among participants on some focus topic. Several challenges arise in helping participants manage their contributions. A semantic index over the participant contributions is used to address these challenges. %N SS-05-01 %U http://www.ai.sri.com/pubs/files/1125.pdf
%0 Conference Proceedings %A Andres Rodriguez, Thomas Boyce, John Lowrance, Eric Yeh %T Angler: Collaboratively Expanding your Cognitive Horizon %B International Conference on Intelligence Analysis %D 2005 %X Angler is a tool to help analysts explore, under-stand, and overcome biases that result from pre-vious experiences and background, and to col-laborate in expanding their joint cognitive vi-sion. Angler utilizes divergent and convergent techniques, such as brainstorming and clustering or voting, to guide a diverse set of intelligence professionals in completing a complex knowl-edge task. The tool helps the group through the process of forming consensus, while preserving and quantifying differing ways of thinking. An-gler provides a Web-based collaborative envi-ronment that allows users distributed by both time and geography to assemble in teams, with the help of a facilitator. %U http://www.ai.sri.com/pubs/files/1233.pdf
%0 Conference Proceedings %A Murray, K. and Harrison, I. and Lowrance, J. and Rodriguez, A. and Thomere, J. and Wolverton, M. %T PHERL: an Emerging Representation Language for Patterns, Hypotheses, and Evidence %B Proceedings of the AAAI Workshop on Link Analysis %D 2005 %X Managing and analyzing information remains an enduring and essential challenge for intelligence professionals. Link analysis tools are beginning to be adopted in the intelligence community as part of the community’s analysis toolset. However, since a wide variety of types of analyses is possible, even within just the link analysis field, no one tool is likely to be sufficient to complete an analysis. Instead, analysts will be required to use a suite of tools to accomplish their tasks. This motivates the need for interchange languages between tools, so they can be used in coordinated workflows to meet the needs of the intelligence analyst. We are developing PHERL, an interchange language for link analysis tools, which is designed to support the sharing of patterns and hypotheses (e.g., pattern-match results). This paper presents our preliminary work and solicits contributions of representation requirements from the link analysis community. %U http://www.ai.sri.com/pubs/files/1145.pdf
%0 Conference Proceedings %A Murray, K. and Lowrance, J. and Appelt, D. and Rodriguez, A. %T Estimating Similarity among Collaboration Contributions %B Third International Conference on Knowledge Capture %D 2005 %X The need for collaboration arises in many activities re-quired for effective problem solving and decision making. We are developing Angler, a web-services tool that sup-ports collaboration among participants on some focus topic. Angler overcomes some common barriers to collaboration by enabling asynchronous and distributed collaboration. Angler supports a collaboration methodology that exploits opportunities afforded by multiple participants each making contributions to the collaboration. One challenge that arises in helping participants manage their contributions and their review of others’ contributions is determining when one contribution is very similar to another contribution. Two very similar contributions may suggest either a need to merge them or to further elaborate one or both of them. Indexes over the participant contributions are used to assess similarity across contributions and address this challenge. The indexes may comprise lexical or ontological informa-tion; the former indexes require fewer resources to deploy but the later appear to support better similarity estimates. %U http://www.ai.sri.com/pubs/files/1179.pdf
%0 Conference Proceedings %A Wolverton, M. and Harrison, I. and Lowrance, J. and Rodriguez, A. and Thomere, J. %T Supporting the Pattern Development Cycle in Intelligence Gathering %B Proceedings of the International Conference on Intelligence Analysis (IA’05) %D 2005 %X To deal with noisy and incomplete data sets, analysts need tools that support an intelligence gathering cycle. In this cycle, the analyst (1) creates an initial pattern corresponding to his information need, (2) retrieves a collection of matching episodes in the data, (3) revises the pattern based on the shortcomings of the matches, and (4) repeats the process until the revised pattern is returning the right data. This paper discusses the cycle through a use case of the Link Analysis Workbench (LAW), a tool for discovering and analyzing situations of interest in large relational data sets.
%0 Conference Proceedings %A Yeh, E. and Boyce, T. and Lowrance, J. and Rodriguez, A. %T A Collaborative Framework for Managing Uncertainty and Cognitive Bias %B The Association of Lisp Users International Lisp Conference %D 2005
%0 Conference Proceedings %A Thomere, J. and Harrison I. and Lowrance J. and Rodriguez A. and Ruspini E. and Wolverton M. %E IEEE %T Helping Intelligence Analysts Detect Threats in Overflowing, Changing and Incomplete Information %B Proceedings of the 2004 IEEE International Conference on Computational Intelligence for Homeland Security and Personal Safety %C S. Giuliano - Venice, Italy %@ 0-7803-8381-8 %I IEEE %P 39-45 %D 2004 %X An important role of intelligence organizations is to be able to identity and predict threats within a vast quantity of imprecise and noisy information. We describe the concept and function of our pattern-matching architecture, LAW (Link Analysis Workbench). This system is based upon two main ideas. The first idea is that both the data and the threats can be described in term of graphs of entities and events linked together with semantic relationships. Therefore, graph-based pattern matching techniques can be used to identify threats. The second idea is that analysts constitute an essential part of the system; LAW is designed to handle a lot of interaction with the user, particularly to help in authoring and revising patterns, by allowing analysts to understand the matching process and results. %U http://www.ai.sri.com/pubs/files/1042.pdf
%0 Journal Article %A Lowrance, J. and Ragoobeer, R. %T Designing a System for Structured Assessment of Compliance Risk %B Proceeding of SRPP Research Conference %D 2004 %X The ability to detect, classify and quantify high-risk compliance patterns is crucial in improving compliance risk assessment and formulating effective enforcement strategies within the IRS. A formal reasoning technique known as structured argumentation has been explored by the IRS LMSB Research to improve compliance risk assessment. The IRS together with SRI International evaluated the use of structured argumentation using SEAS technology, towards enabling a more systematic and innovative approach towards assessing complex compliance issues. Technical experts, auditors, attorneys, researchers and managers can work collaboratively to unravel and respond to high-risk compliance patterns more rapidly and preserve this knowledge to be shared and referenced later. %U http://www.ai.sri.com/pubs/files/1052.pdf
%0 Generic %A Wolverton, M. and Berry, P. and Harrison, I. and Lowrance, J. and Morley, D. and Rodriguez, A. and Ruspini, E. and Thomere, J. %T LAW: A Workbench for Approximate Pattern Matching in Relational Data %D 2003 %X Pattern matching for intelligence organizations is a challenging problem. The data sets are large and noisy, and there is a flexible and constantly changing notion of what constitutes a match. We are developing the Link Analysis Workbench (LAW) to assist an expert user in the intelligence community in creating and maintaining patterns, matching those patterns against a large collection of relational data, and manipulating partial results. This paper describes two key facets of the LAW system: (1) a pattern-matching module based on a graph edit distance metric, and (2) a system architecture that supports the integration and tasking of multiple pattern matching modules based on their capabilities and the specific problem at hand. %O in The Fifteenth Innovative Applications of Artificial Intelligence Conference (IAAI-03) %U http://www.ai.sri.com/pubs/files/931.pdf
%0 Conference Proceedings %A Mishra, S. and Rodriguez, A. and Eriksen, M. and Chaudhri, V. and Lowrance, J. and Murray, K. and Thomere, J. %T Lightweight solutions for user interfaces over the WWW %B Proceedings of the International Lisp Conference %D 2002 %U http://www.ai.sri.com/pubs/files/920.pdf
%0 Generic %A Lowrance, J., Harrison, I., and Rodriguez, A. %T SEAS Help System %7 Version 5.1 %C 333 Ravenswood Avel, Menlo Park, CA 94025 %D 2001 %X This constitutes the user manual for SEAS, the Structured Evidential Argumentation System.
%0 Journal Article %A Lowrance, John D., Harrison, Ian W., and Rodriguez, Andres C. %T Capturing Analytic Thought %B Proceeding of the First International Conference on Knowledge Capture %P 84-91 %D 2001 %X The survival of an enterprise often rests upon its ability to make correct and timely decisions, despite the complexity and uncertainty of the environment. Because of the difficulty of employing and scaling formal methods in this context, decision makers typically resort to informal methods, sacrificing structure and rigor. We are developing a new methodology that retains the ease of use, the familiarity, and (some of) the free-form nature of informal methods, while benefiting from the rigor, structure, and potential for automation characteristic of formal methods. Our approach records analysts’ thinking in a corporate knowledge base consisting of structured arguments. The foundation of this knowledge base is an ontology of arguments that includes two main types of formal objects: argument templates and arguments. An argument template records an analytic method as a hierarchically structured set of interrelated questions, and an argument instantiates an argument template by answering the questions posed relative to a specific situation. This methodology emphasizes the use of simple inference structures as the foundation of its argument templates, making it possible for analysts to independently author new templates. When authoring an argument template, the analyst can choose to embed discovery tools, which are recommended methods of acquiring information pertaining to the questions posed. An analyst wanting to record an argument selects an appropriate template, uses the discovery tools to retrieve potentially relevant information, selects that information to retain as evidence and records its relevance, answers the questions, and records the rationale for the answers. The result is a recorded line of reasoning that breaks down the problem, bottoming out at the documents and other forms of information that were used as evidence to support the answers. The resulting collection of arguments and templates constitutes a corporate memory of analytic thought that can be directly exploited by analysts or automated methods. %O Available from the ACM Portal.
%0 Conference Proceedings %A Lowrance, John D. and Harrison, Ian W. and Rodriguez, Andres C. %T Structured Argumentation for Analysis %B Proceedings of the 12th International Conference on Systems Research, Informatics, and Cybernetics: Focus Symposia on Advances in Computer-Based and Web-Based Collaborative Systems %C Baden-Baden, Germany %P 47-57 %D 2000 %X The survival of an enterprise often rests upon its ability to make correct and timely decisions, despite the complexity and uncertainty of the environment. Because of the difficulty of employing formal methods in this context, decision makers typically resort to informal methods, sacrificing structure and rigor. We are developing a new methodology that retains the ease-of-use, familiarity, and (some of) the free-form nature of informal methods, while benefiting from the rigor, structure, and potential for automation characteristic of formal methods. Our approach aims to foster thoughtful and timely analysis through the introduction of structure, and collaboration through access to current and past analytic results. By recording analysts’ thinking in structured arguments, the results are more comprehensive, more readily understood, and more easily compared. By providing access to a corporate memory of analytic methods and results, analysts can work together on common arguments and leverage historical results. The structured argumentation methodology encourages a careful analysis, by reminding the analyst of the full spectrum of indicators to be considered. It also eases argument comprehension by allowing the analyst or decision maker to drill down along the component lines of reasoning to discover the basis and rationale of others’ arguments, and it invites and facilitates argument comparison by framing arguments within common structures. %U http://www.ai.sri.com/pubs/files/434.pdf
%0 Journal Article %A Tao, K. M., Abileah, R., and Lowrance, J. D. %T Multiple-Target Tracking and Data Fusion via Probabilistic MappingFusion via Probabilistic Mapping %B Proc. 2000 MSS National Symposium on Sensor and Data Fusion %D 2000 %X A new approach is taken to address the various aspects of the multi-sensor, multi-target tracking (MTT) problem in dense and noisy environments. Instead of fixing the trackers on the potential targets as the conventional tracking algorithms do, this new approach is fundamentally different in that an array of parallel-distributed “trackers” is laid in the search space. The difficult data-track association problem that has challenged the conventional trackers becomes a nonissue with this new approach. By partitioning the search space into “cells,” this new approach, called PMAP (probabilistic mapping), dynamically calculates the spatial probability distribution of targets in the search space via Bayesian updates. The distribution is spread at each time step, following a fairly general Markov-chain target motion model, to become the prior probabilities of the next scan. This framework can effectively handle data from multiple sensors and incorporate contextual information, such as terrain and weather, by performing a form of evidential reasoning. Used as a pre-filtering device, the PMAP is shown to remove noiselike false alarms effectively, while keeping the target dropout rate very low. This gives the downstream track linker a much easier job to perform. A related benefit is that with PMAP it is now possible to lower the detection threshold and to enjoy high probability of detection and low probability of false alarm at the same time, thereby improving overall tracking performance. The feasibility of using PMAP to track specific targets in an end-game scenario is also demonstrated. Both real and simulated data are used to illustrate the PMAP performance. The PMAP algorithm is parallel distributed in nature; for serial computer implementation, fast algorithms have been developed. Some related applications based on the PMAP approach, including a spatial–temporal sensor data fusion application and a gray-scale video sequence stacking application, are also discussed. %U http://www.ai.sri.com/pubs/files/872.pdf
%0 Journal Article %A Tao, K.M., Abileah, R. and Lowrance, J.D. %T Multiple-Target Tracking in Dense, Noisy Environments: A Probabilistic Mapping Perspective %B Proc. SPIE: Signal and Data Processing of Small Targets 2000 %P 474-485 %V 4048 %D 2000 %X A new approach is taken to address the various aspects of the multiple-target tracking (MTT) problem in dense and noisy environments. Instead of fixing the trackers on the potential targets as the conventional tracking algorithms do, this new approach is fundamentally different in that an array of parallel-distributed ‘trackers’ is laid in the search space. The difficult data-track association problem that has challenged the conventional trackers becomes a non-issue with this new approach. By partitioning the search space into ‘cells,’ this new approach, called PMAP (probabilistic Mapping), dynamically calculates the spatial probability distribution of targets in the search space via Bayesian updates. The distribution is spread at each time step, following some fairly general Markov-chain target motion model, to become the prior probabilities of the next scan. This framework can effectively handle data from multiple sensors and incorporates contextual information, such as terrain and weather, by performing a form of Evidential Reasoning. Used as a pre-filtering device, the PMAP is shown to remove noise-like false alarms effectively, while keeping target dropout rate very low. This gives the downstream track ‘linker’ a much easier job to perform. A related benefit is that with PMAP it is now possible to lower the detection threshold and to enjoy high Probability of Detection and low Probability of False Alarm at the same time, thereby improving overall tracking performance. The feasibility of using PMAP to track specific targets in an end-game scenario is also discussed. Both real and simulated data are used to illustrate the PMAP performance. Some related applications based on the PMAP approach, including a spatial-temporal sensor data fusion application, are mentioned.
%0 Report %@ Ontology %A Chaudhri, Vinay K. and Lowrance, John D. and Stickel, Mark E. and Thomere, Jerome F. and Wadlinger, Richard J. %T Ontology Construction Toolkit %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 2000 %U http://www.ai.sri.com/pubs/files/783.pdf
%0 Report %A Tao, K. Mike and Abileah, Ronald and Ravichandran, Gopalan and McPherrin, David L. and Lowrance, John D. and Ruspini, Enrique H. and Milanfar, Peyman %T Multiple-Target Tracking in Dense, Noisy Environments %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1999
%0 Conference Proceedings %A Paley, S.M., Lowrance, J.D., and Karp, P.D. %T A Generic Knowledge-Base Browser and Editor %B Proceedings of the 1997 National Conference on Artificial Intelligence %D 1997 %U http://www.ai.sri.com/pubs/files/895.ps
%0 Generic %A Garvey, Thomas D. and Lowrance, John D. and Fischler, Martin A. %E Luo, Ren C. and Kay, Michael G. %T An Inference Technique for Integrating Knowledge from Disparate Sources %B Multisensor Integration and Fusion for Intelligenct Machines and Systems %C Norwood, New Jersey %I Ablex Publishing Corporation. %P 309-325 %D 1995 %X This paper introduces a formal method for integrating knowledge derived from a variety of sources for use in ``perceptual reasoning.’’ The formalism is based on the ``evidential propositional calculus.’’ a derivative of Shafer’s mathematical theory of evidence [Shafer 1976]. It is more general than either a Boolean or Bayesian approach, providing for Boolean and Bayesian inferencing when the appropriate information is available. In this formalism, the likelihood of a proposition A is represented as a subinterval, [s(A),p(A)], of the unit interval [0,1]. The evidential support for the proposition A is represented by s(A), while p(A) represents its degree of plausibility; p(A) can also be interpreted as the degree to which one fails to doubt A, p(A) being equal to one minus the evidential support for ~A. This paper describes how evidential information, furnished by a knowledge source in the form of a probability ``mass’’ distribution, can be converted to this interval representation; how, through a set of inference rules for computing intervals of dependent propositions, this information can be extrapolated from those propositions it directly bears upon, to those it indirectly bears upon; and how multiple bodies of evidential information can be pooled. A sample application of this approach, modeling the operation of a collection of sensors (a particular type of knowledge source), illustrates these techniques.
%0 Journal Article %A Wilkins, David E. and Myers, Karen L. and Lowrance, John D. and Wesley, Leonard P. %T Planning and Reacting in Uncertain and Dynamic Environments %B Journal of Experimental and Theoretical AI %I Taylor %P 197-227 %V 7 %D 1995 %X Agents situated in dynamic and uncertain environments require several capabilities for successful operation. Such agents must monitor the world and respond appropriately to important events. The agents should be able to accept goals, synthesize complex plans for achieving those goals, and execute the plans while continuing to be responsive to changes in the world. As events render some current activities obsolete, the agents should be able to modify their plans while continuing activities unaffected by those events. The CYPRESS system is a domain-independent framework for defining persistent agents with this full range of behavior. CYPRESS has been used for several demanding applications, including military operations, real-time tracking, and fault diagnosis. %N 1 %U http://www.ai.sri.com/pubs/files/262.ps
%0 Journal Article %A Stokke, Per R. and Boyce, Thomas A. and Lowrance, John D. and William K. Ralston, Jr. %T Industrial Project Monitoring with Evidential Reasoning %B Nordic Advanced Information Technology Magazine %P 18-27 %V 8 %D 1994 %X Current project management systems share a common weakness. Although all provide ample information about the current status of a project, none provide meaningful information about the likely outcome of that project. Integrating information about the wide range of factors that affect project success and using that information to monitor and take early corrective action form the bases of the Project Early Warning System (PEWS). The system identifies problems and developments that might lead to deviations from planned project outcomes and does so at such an early stage that effective corrective action can still be taken. PEWS combines a proven project reporting methodology with the latest artificial intelligence techniques such as evidential reasoning. Together, they ensure the successful outcome of large projects. The system encourages objective assessment and reporting by project leaders, while providing upper management with a clear and concise report that pinpoints aspects of projects in the company's portfolio. %N 1
%0 Journal Article %A Karp, P. D. and Lowrance, J. D. and Strat, T. M. and Wilkins, D. E. %T The Grasper-CL Graph Management System %B LISP and Symbolic Computation %P 245–282 %V 7 %D 1994 %X Graphs are virtually ubiquitous in programming applications. Moreover, graph-structured information is especially prevalent in AI applications. We can enhance programs that manipulate graph-structured information by providing these programs with graphical user interfaces that draw graphs, and that allow users to interact with drawings of graph nodes and edges. Grasper-CL is a Common Lisp system for manipulating and displaying graphs. Grasper-CL defines a graph abstract datatype and an extensive set of associated operations for creating, modifying and interrogating graphs, and for saving them persistently. The system draws graphs using CLIM (the Common Lisp Interface Manager), and can create postscript renditions of its drawings. Grasper-CL supports a wide variety of graphic styles for drawing graph nodes and edges. The system includes several different automatic graph layout algorithms, such as for circular and tree layout; it also supports full interactive manipulation of graph drawings. Finally, the system provides facilities for building graph-based user interfaces for application programs, which have been used in conjunction with the Sipe planner, the Gister evidential reasoner, a scheduler for the Hubble Space Telescope, and the EcoCyc encyclopedia of biochemical pathways. A number of groups within the AIC and SRI are using the Grasper-CL system in a variety of projects. This talk will describe the system in detail for people who wish to understand its capabilities better or who are thinking of using it for other projects. This talk is also an opportunity for the audience to shape the future directions of the system: What additional capabilities should be added? Would users like more direct input in how the system evolves? Should we attempt to find funding for further development of the system and research on such issues as graph layout algorithms? %U http://www.ai.sri.com/pubs/files/237.ps.Z
%0 Generic %A Lowrance, John D. %T Evidential Reasoning with Gister-CL: A Manual %C 333 Ravenswood Avenue, Menlo Park, CA %D 1994 %X This document is designed to serve as a self-contained introduction to evidential reasoning and Gister-CL. Evidential reasoning is a collection of techniques for automated reasoning from evidence; Gister-CL is an application independent implementation of these techniques. As such, Gister-CL might serve either as the foundation for application specific implementations of evidential reasoning or as a basis for research in uncertain reasoning. Both evidential reasoning and Gister-CL are undergoing further development. Therefore, this document will be periodically updated.
%0 Journal Article %A Stokke, Per R. and Boyce, Thomas A. and Lowrance, John D. and William K. Ralston, Jr. %T Evidential Reasoning and Project Early Warning Systems %B Research and Technology Management %D 1994 %X Current project management systems share a common weakness. Although all provide ample information about the current status of a project, none provide meaningful information about the likely outcome of that project. Integrating information about the wide range of factors that affect project success and using that information to monitor and take early corrective action form the bases of the Project Early Warning System (PEWS). The system identifies problems and developments that might lead to deviations from planned project outcomes and does so at such an early stage that effective corrective action can still be taken. PEWS combines a proven project reporting methodology with the latest artificial intelligence techniques such as evidential reasoning. Together, they ensure the successful outcome of large projects. The system encourages objective assessment and reporting by project leaders, while providing upper management with a clear and concise report that pinpoints aspects of projects in the company's portfolio.
%0 Journal Article %A Ruspini, Enrique H. and Lowrance, John D. and Strat, Thomas M. %E Bezdek, James C. %T Understanding Evidential Reasoning %B International Journal of Approximate Reasoning %I North-Holland %P 401-424 %V 6 %D 1992 %X We address recent criticisms of evidential reasoning, an approach to the analysis of imprecise and uncertain information that is based on the Dempster-Shafer calculus of evidence. We show that evidential reasoning can be interpreted in terms of classical probability theory and that the Dempster-Shafer calculus of evidence may be considered to be a form of generalized probabilistic reasoning based on the representation of probabilistic ignorance by intervals of possible values. In particular, we emphasize that it is not necessary to resort to nonprobabilistic or subjectivist explanations to justify the validity of the approach. We answer conceptual criticisms of evidential reasoning primarily on the basis of the criticism's confusion between the current state of development of the theory---mainly theoretical limitations in the treatment of conditional information---and it potential usefulness in treating a wide variety of uncertainty analysis problems. Similarly, we indicate that the supposed lack of decision-support schemes of generalized probability approaches is not a theoretical handicap but rather an indication of basic informational shortcoming that is a desirable asset of any formal approximate reasoning approach. We also point to potential shortcomings of the underlying representation scheme to treat general probabilistic reasoning problems. We also consider methodological criticisms of the approach, focusing primarily on the alleged counterintuitive nature of Dempster's combination formula, showing that such results are the result of its misapplication. We also address issues of complexity and validity of scope of the calculus of evidence. %N 3
%0 Report %@ SRI Contra %A Lang, Ruth E. and Lowrance, John D. and Cohen, Philip R. and Lunt, Teresa F. %T A Study in the Application of Artificial Intelligence Technology to the DOD Directory %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1992
%0 Conference Proceedings %A Gaon, David and Lang, Ruth E. and Lowrance, John D. and Cohen, Philip R. %T Application of Artificial Intelligence to the DOD Directory %B Proceedings of MILCOM '92 %D 1992 %X The Department of Defense (DoD) is planning for the implementation of a DoD Directory capability based on CCITT Recommendations X.500-X.521, which define the Data Communication Networks Directory. The functional and operational requirements that define the DoD Directory will yield a system with a significant level of complexity. Problems and barriers which impede progress toward the envisioned DoD Directory service exist. This paper describes these problems and the artificial intelligence-based approach developed to solve and reduce them in order to achieve a usable, capable, secure, and manageable DoD Directory service.
%0 Report %@ 521 %A Peter D. Karp, John D. Lowrance, Thomas M and Wilkins, David E. %T The Grasper-CL Graph Management System %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1992 %X Grasper-CL [5] is a COMMON LISP system for manipulating anddisplaying graphs, and for building graph-based user interfaces forapplication programs. The system represents a significant advance overprevious COMMON LISP graphers because each level of the Grasper-CLarchitecture—from the core graph data structures to the interactivedisplay module—has been fully developed and articulated, and isaccessible to application programmers. We call this system organizationan open architecture. In our experience, several different classes ofgraph-based user interfaces exist. For example, one class producesstatic drawings of graphs, whereas another class requires extensive userinteraction with graph drawings. The open architecture of Grasper-CLsupports the development of all classes of interfaces, whereas previousgraphers support only one or two classes of interfaces. Grasper-CLgraphics operations are implemented using CLIM, the COMMON LISPInterface Manager. %U http://www.ai.sri.com/pubs/files/454.pdf
%0 Report %@ SRI Contra %A Lowrance, John D. and Strat, Thomas M. and Wesley, Leonard P. and Garvey, Thomas D. and Ruspini, Enrique H. and Wilkins, David E. %T The Theory, Implementation, and Practice of Evidential Reasoning %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1991 %X Evidential Reasoning (ER) is a body of techniques for automated reasoning from evidence that is based upon the mathematics of Dempster-Shafer belief functions. The emphasis of this project was twofold: to broaden and solidify the theoretical basis of ER, and to facilitate the transfer of the intellectual technology embodied in ER. As part of our theoretical effort, we established a sound semantics for Dempster-Shafer belief functions, deriving the Dempster-Shafer axioms based upon epistemic logic; we derived all ER operations from these same Dempster-Shafer axioms; we established ER as a generalization of both logical and Bayesian probabilistic reasoning; we identified the conditions under which the computational complexity of ER belief networks can be reduced; we developed a theoretically justified means of propagating all information throughout an evidential analysis, determining the global impact on all probabilistically dependent random variables; we incorporated decision theoretic concepts into ER, including sensitivity analysis and decision trees. To facilitate the transfer of this technology, we developed intuitive graphical structures for representing and manipulating evidential knowledge, thereby, substantially reducing the time required to compile and organize the knowledge for a new application domain, and we developed a set of canonical ER examples covering a range of application domains, including underwater vehicle tracking, antiair threat identification, medical diagnosis, and robot vehicle navigation.
%0 Report %@ 501 %A Ruspini, Enrique and Lowrance, John D. and Strat, Thomas M. %T Understanding Evidential Reasoning %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1990 %K Evidential Reasoning!Understanding, Reasoning!Uncertain|seeEvidential Reasoning %X We address recent criticisms of evidential reasoning, an approach to the analysis of imprecise and uncertain information that is based on the Dempster-Shafer calculus of evidence. We show that evidential reasoning can be interpreted in terms of classical probability theory and that the Dempster-Shafer calculus of evidence may be considered to be a form of generalized probabilities reasoning based on the representation of probabilistic ignorance by intervals of possible values. In particular, we emphasize that it is not necessary to resort to nonprobabilistic or subjectivist explanations to justify the validity of the approach. %U http://www.ai.sri.com/pubs/files/473.pdf
%0 Generic %A Lowrance, John D. and Wilkins, David E. %E Sycara, Katia P. %T Plan Evaluation under Uncertainity %B Proceedings of the Workshop on Innovative Approaches to Planning, Scheduling and Control %I Morgan Kaufmann Publishers Inc., San Mateo, CA %P 439-449 %D 1990 %U http://www.ai.sri.com/pubs/files/256.ps
%0 Generic %A Lowrance, John D. and Garvey, Thomas D. and Strat, Thomas M. %E Shafer, Glenn and Pearl, Judea %T A Framework for Evidential-Reasoning Systems %B Uncertain Reasoning %C San Mateo, CA %I Morgan Kaufman Publishers, Inc. %P 611-618 %D 1990 %X Evidential reasoning is a body of techniques that supports automated reasoning from evidence. It is based upon the Dempster-Shafer theory of belief functions. Both the formal basis and a framework for the implementation of automated reasoning systems based upon these techniques are presented. The formal and practical approaches are divided into four parts (1) specifying a set of distinct propositional spaces, each of which delimits a set of possible world situations (2) specifying the interrelationships among these propositional spaces (3) representing bodies of evidence as belief distributions over these propositional spaces and (4) establishing paths for the bodies of evidence to move through these propositional spaces by means of evidential operations, eventually converging on spaces where the target questions can be answered.
%0 Journal Article %A Strat, Thomas M. and Lowrance, John D. %E Bezdek, James C. %T Explaining Evidential Analyses %B International Journal of Approximate Reasoning %I North-Holland %P 299-353 %V 3 %D 1989 %X One of the most highly touted virtues of knowledge-based expert systems is their ability to construct explanations in for their lines of reasoning. However, there is a basic difficulty in generating explanations in espert systems that reason under uncertainty using numeric measures. In particular, systems based upon evidential reasoning using the theory of belief functions have lacked all but the most rudimentary facilities for explaining their conclusions. In this paper we review the process whereby other expert system technologies produce explanations, and present a methodology for augmenting an evidential-reasoning system with a versatile explanation facility. The method, which is based on sensitivity analysis, has been implemented, and several examples of its use are described. %N 4
%0 Conference Proceedings %A Lowrance, John D. %T Automating Multisource Data Analysis %B Proceedings of the International Lithosphere Project Research Conference on Advanced Data Integration in Mineral and Energy Resource Studies %C Sotogrande, Spain %D 1988 %X Over the past eight years, the Artificial Intelligence Center at SRI International has been developing new technology to address the problem of automated information management within real-world contexts. The result of this work is a body of techniques for automated reasoning from evidence, called evidential reasoning. These techniques emphasize the ability to reason from information that is uncertain, incomplete, and inaccurate. To support this line of research we developed Gister, a domain-independent evidential-reasoning system. Gister provides both a formal basis and an implementation framework for automated argument construction based upon evidential-reasoning techniques. It supports the construction, modification, and interrogation of lines of reasoning that argue from multiple bodies of evidence toward probabilistic conclusions. We believe that the current Gister system could offer significant aid to analysts performing a wide range of multisource data analysis tasks. Furthermore, Gister provides the necessary basis for generalized argument construction, the central requirement of real-world analysis.
%0 Journal Article %A Lowrance, John D. %T Automated Argument Construction %B Journal of Statistical Planning and Inference %P 369-387 %V 20 %D 1988
%0 Report %@ 430 %A Strat, Thomas M. and Lowrance, John D. %T Explaining Evidential Analyses %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1988 %K Evidential Reasoning!Explaining, Reasoning!Evidential|seeEvidential Reasoning %X One of the most highly touted virtues of knowledge-based expert systems is their ability to construct explanations for their lines of reasoning. However, there is a basic difficulty in generating explanations in expert systems that reason under uncertainty using numeric measures. In particular, systems based upon evidential reasoning using the theory of belief functions have lacked all but the most rudimentary facilities for explaining their conclusions. In this paper we review the process whereby other expert system technologies produce explanations, and present a methodology for augmenting an evidential-reasoning system with a versatile explanation facility. The method, which is based on sensitivity analysis, has been implemented, and several examples of its use are described. %U http://www.ai.sri.com/pubs/files/536.pdf
%0 Conference Proceedings %A Lowrance, John D. and Garvey, Thomas D. %T Automating Argument Construction for Intelligence Analysis %B Proceedings of the Military Computing Conference %C Anaheim, CA %D 1987
%0 Generic %A Lowrance, John D. %T Evidential Reasoning with Gister: A Manual %C 333 Ravenswood Avenue, Menlo Park, CA %D 1987
%0 Conference Proceedings %A Lowrance, John D. %T Automating Argument Construction %B Proceedings of the Workshop on Assessing Uncertainty (November 13-14, 1986) %C Department of Statistics, Stanford University, Stanford, CA %D 1987 %X Over the past five years the Artificial Intelligence Center at SRI has been developing a new technology to address the problem of automated information management within real-world contexts. The result of this work is a body of techniques for automated reasoning from evidence that we call evidential reasoning. The techniques are based upon the mathematics of belief functions developed by Dempster and Shafer and have been successfully applied to a variety of problems including computer vision, multisensor integration, and intelligence analysis. We have developed both a formal basis and a framework for implementating automated reasoning systems based upon these techniques. Both the formal and practical approach can be divided into four parts: (1) specifying a set of distinct propositional spaces, (2) specifying the interrelationships among these spaces, (3) representing bodies of evidence as belief distributions, and (4) establishing paths for the bodies of evidence to move through these spaces by means of evidential operations, eventually converging on spaces where the target questions can be answered. These steps specify a means for arguing from multiple bodies of evidence toward a particular (probabilistic) conclusion. Argument construction is the process by which such evidential analyses are constructed and is the analogue of constructing proof trees in a logical context. This technology features the ability to reason from uncertain, incomplete, and occasionally inaccurate information based upon seven evidential operations: fusion, discounting, translating, projection, summarization, interpretation, and gisting. These operation are theoretically sound but have intuitive appeal as well. In implementing this formal approach, we have found that evidential arguments can be represented as graphs. To support the construction, modification, and interrogation of evidential arguments, we have developed Gister. Gister provides an interactive, menu-driven, graphical interface that allows these graphical structures to be easily manipulated. Our goal is to provide effective automated aids to domain experts for argument construction. Gister represents our first attempt at such an aid.
%0 Generic %A Lowrance, John D. %T Grasper II Reference Manual %C 333 Ravenswood Avenue, Menlo Park, CA %D 1987
%0 Report %@ 416 %A Lowrance, John D. %T Automating Argument Construction %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1986 %K Evidential Reasoning!Formal basis, GISTER, Evidential Reasoning!GISTER %X Over the past five years the Artificial Intelligence Center at SRI has been developing a new technology to address the problem of automated information management within real-world contexts. The result of this work is a body of techniques for automated reasoning from evidence that we call evidential reasoning. The techniques are based upon the mathematics of belief functions developed by Dempster and Shafer and have been successfully applied to a variety of problems including computer vision, multisensor integration, and intelligence analysis. We have developed both a formal basis and a framework for implementing automated reasoning systems based upon these techniques. Both the formal and practical approach can be divided into four parts: (1)~specifying a set of distinct propositional spaces, (2)~specifying the interrelationships among these spaces, (3)~representing bodies of evidence as belief distributions, and (4)~establishing paths for the bodies of evidence to move through these spaces by means of evidential operations, eventually converging on spaces where the target questions can be answered. These steps specify a means for arguing from multiple bodies of evidence toward a particular (probabilistic) conclusion. Argument construction is the process by which such evidential analyses are constructed and is the analogue of constructing proof trees in a logical context. This technology features the ability to reason from uncertain, incomplete, and occasionally inaccurate information based upon seven evidential operations: fusion, discounting, translating, projection, summarization, interpretation, and gisting. These operations are theoretically sound but have intuitive appeal as well. In implementing this formal approach, we have found that evidential arguments can be represented as graphs. To support the construction, modification, and interrogation of evidential arguments, we have developed Gister. Gister provides an interactive, menu-driven, graphical interface that allows these graphical structures to be easily manipulated. Our goal is to provide effective automated aids to domain experts for argument construction. Gister represents our first attempt at such an aid. %U http://www.ai.sri.com/pubs/files/549.pdf
%0 Conference Proceedings %A Lowrance, John D. and Garvey, Thomas D. and Strat, Thomas M. %T A Framework for Evidential-Reasoning Systems %B Proceedings of the National Conference on Artificial Intelligence %C Menlo Park, CA %P 896-903 %D 1986 %X Evidential reasoning is a body of techniques that supports automated reasoning from evidence. It is based upon the Dempster-Shafer theory of belief functions. Both the formal basis and a framework for the implementation of automated reasoning systems based upon these techniques are presented. The formal and practical approaches are divided into four parts (1) specifying a set of distinct propositional spaces, each of which delimits a set of possible world situations (2) specifying the interrelationships among these propositional spaces (3) representing bodies of evidence as belief distributions over these propositional spaces and (4) establishing paths for the bodies of evidence to move through these propositional spaces by means of evidential operations, eventually converging on spaces where the target questions can be answered.
%0 Report %@ SRI Contra %A Lowrance, John D. and Strat, Thomas M. and Garvey, Thomas D. %T Application of Artificial Intelligence Techniques to Naval Intelligence Analysis %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1986 %X Toward the goal of providing automated aides for intelligence analysis, SRI, in cooperation with DARPA and NAVALEX, developed a system called Navint. Navint is an interactive system that aids naval intelligence analysts in estimating the location and activity of surface vessels, based upon a variety of intelligence reports from multiple sources. It focuses on ship-by-ship estimates, using these as the basis for a fleet wide summary. The analyst interacts with the system through a set of display windows. The primary window is a ledger or ``timeline'' that is used to record and organize the daily intelligence data and products for each ship. The mode of interaction is modeled after electronic spread-sheet programs. As the analyst fills in the (electronic) form with the available information, Navint automatically updates related entries. A different view of this same information is provided through ``map'' windows. These display the presumed routes of individual fleet elements overlaid on maps of the appropriate areas. Another display tracks the flow of intelligence information through evidential-reasoning operations (e.g., discounting, fusing, translation, and projection). When a new report is entered by the analyst, a new node is created in this ``analysis'' display to represent the newly acquired information. The application of evidential-reasoning operations spawns new nodes related to preexisting ones. When new nodes are created, labeled links are created between these nodes to trace the heritage of each resulting body of information. Thus, explanations of selected conclusions can be generated by tracing back along these links to find the supporting reports and to determine their relative roles.
%0 Generic %A Garvey, Thomas D. and Lowrance, John D. and Fischler, Martin A. %E Brachman, Ronald J. and Levesque, Hector J. %T An Inference Technique for Integrating Knowledge from Disparate Sources %B Readings in Knowledge Representation %C San Mateo, CA %I Morgan Kaufman Publishers, Inc. %P 457-464 %D 1985 %X This paper introduces a formal method for integrating knowledge derived from a variety of sources for use in ``perceptual reasoning.’’ The formalism is based on the ``evidential propositional calculus.’’ a derivative of Shafer’s mathematical theory of evidence [Shafer 1976]. It is more general than either a Boolean or Bayesian approach, providing for Boolean and Bayesian inferencing when the appropriate information is available. In this formalism, the likelihood of a proposition A is represented as a subinterval, [s(A),p(A)], of the unit interval [0,1]. The evidential support for the proposition A is represented by s(A), while p(A) represents its degree of plausibility; p(A) can also be interpreted as the degree to which one fails to doubt A, p(A) being equal to one minus the evidential support for ~A. This paper describes how evidential information, furnished by a knowledge source in the form of a probability ``mass’’ distribution, can be converted to this interval representation; how, through a set of inference rules for computing intervals of dependent propositions, this information can be extrapolated from those propositions it directly bears upon, to those it indirectly bears upon; and how multiple bodies of evidential information can be pooled. A sample application of this approach, modeling the operation of a collection of sensors (a particular type of knowledge source), illustrates these techniques.
%0 Report %@ SRI Contra %A Garvey, Thomas D. and Lowrance, John D. %T Issues and Approaches to Planning Under Uncertainty %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1984
%0 Journal Article %A Garvey, Thomas D. and Lowrance, John D. %T An AI Approach to Information Fusion %B Journal of Electronic Defense %P 31-41 %D 1984
%0 Report %@ 324 %A Wesley, Leonard P. and Lowrance, John D. and Garvey, Thomas D. %T Reasoning About Control: An Evidential Approach %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1984 %K Evidential Reasoning!About control %X Expert systems that operate in complex domains are continually confronted with the problem of deciding what to do next. Being able to reach a decision requires, in part, having the capacity to ``reason’’ about a set of alternative actions. It has been argued that expert systems must reason from evidential information–i.e., uncertain, incomplete, and occasionally inaccurate information [Low82a]. As a consequence, a model for reasoning about control must be capable of performing several tasks: to combine the evidential information that is generically distinct and from disparate sources; to overcome minor inaccuracies in the evidential information that is needed to reach a decision; to reason about what additional evidential information is required; to explain the actions taken (based on such information) by the system. These are a few of the formidable control problems that remain largely unsolved [Bar82]. If expert systems are to improve their performance significantly, they must utilize increasingly sophisticated and general models for dealing with the evidential information required for reasoning about their behavior. To this end we present an alternative evidentially-based approach to reasoning about control that has several advantages over existing techniques. It enables us to reason from limited and imperfect information; to partition bodies of meta- and domain-knowledge into modular components; and to order potential actions flexibly by allowing any number of constraints (i.e., control strategies) to be imposed over a set of alternative actions. Furthermore, because it can be used for reasoning about the expenditure of additional resources to obtain the evidential information needed as a basis for choosing among alternatives, this approach can be employed recursively. %U http://www.ai.sri.com/pubs/files/628.pdf
%0 Report %@ 318 %A Lowrance, John D. and Garvey, Thomas D. %T An AI Approach to the Integration of Information %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1984 %U http://www.ai.sri.com/pubs/files/768.pdf
%0 Report %@ 318 %A Garvey, Thomas D. and Lowrance, John D. %T An Ai Approach To Information Fusion %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1983 %K Evidential Reasoning!Information Fusion %X This paper discusses the use of selected artificial intelligence (AI) techniques for integrating multisource information in order to develop an understanding of an ongoing situation. The approach takes an active, ``top-down'' view of the task, projecting a situation description forward in time, determining gaps in the current model, and tasking sensors to acquire data to fill the gaps. Information derived from tasked sensors and other sources is combined using new, non-Bayesian inference techniques. This active approach seems critical to solve the problems posed by the low emission signatures anticipated for near-future threats. Simulation experiments lead to the conclusion that the utility of ESM system operation in future conflicts will depend on how effectively onboard sensing resources are managed by the system. The view of AI that will underly the discussion is that of a technology attempting to extend automation capabilities from the current ``replace the human's hands'' approach to that of replacing or augmenting the human's cognitive and perceptual capabilities. Technology transfer issues discussed in the presentation are the primary motivation for highlighting this view. The paper will conclude with a discussion of unresolved problems associated with the introduction of AI technology into real world military systems.
%0 Report %@ 307 %A Lowrance, John D. and Garvey, Thomas D. %T Evidential Reasoning: An Implementation For Multisensor Integration %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1983 %K Evidential Reasoning!Multisensor integration %X Reasoning from uncertain, incomplete, and sometimes inaccurate information is necessary whenever any system is to interact in an intelligent way with its environment. This follows directly from the fact that understanding the world is possible only by perceiving it through a set of knowledge sources that provide partially processed sensory information. Because of the limited capabilities of any sensor, the information is inherently ``evidential.’’ That is, perceptual information is not readily captured in terms of simple truths and falsities or in terms of probabilistic estimates, when the appropriate statistical data are lacking. Therefore, neither logical nor standard probabilistic reasoning techniques are uniformly applicable in this context. %U http://www.ai.sri.com/pubs/files/643.pdf
%0 Report %@ SRI Contra %A Garvey, Thomas D. and Lowrance, John D. %T Machine Intelligence for Electronic Warfare Applications %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1983 %X The primary objective of these two-year efforts was to extend the available AI-based techniques for drawing conclusions from multisource information. The specific problem addressed was that of integrating sensor data with prior knowledge in order to update and maintain a current threat order of battle for a penetrating aircraft. The goal was to provide an automated capability to replace the judgemental capabilities of a human operator in interpreting the returns from his on-board sensors to best assess the antiair threats arrayed against him. The key problems were to find a way of representing and reasoning about the wide variety of knowledge sources a human operator would normally invoke in performing this task. A secondary objective was to demonstrate the effectiveness of these techniques by simulating the interaction of sensors such as radar warning systems and an optical augmentation device. The results of this simulation demonstrated that the optical augmentation device could be automatically cued by the radar warning receiver and resulted in a decrease in the overall false alarm rate and an increase in the number of successful detections. The work also resulted in the development and demonstrated the utility of a new collection of inference techniques labeled evidential reasoning. These techniques represented a significant advance in the state of the art of methods for reasoning with uncertain knowledge.
%0 Report %A Lowrance, John D. and Garvey, Thomas D. %T Evidential Reasoning: An Approach to the Simulation of a Weapons Operation Center %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1983 %X This one year study defined the role of evidential reasoning for simulating activities in a Soviet weapons operation center, and resulted in a specification for the development of a WOC simulation, based on AI technology.
%0 Conference Proceedings %A Lowrance, John D. and Garvey, Thomas D. %T Evidential Reasoning: A Developing Concept %B Proceedings of the Internation Conference on Cybernetics and Society %P 6-9 %D 1982 %X One common feature of most knowledge-based expert systems is that they must draw conclusions on the basis of evidential information. Yet there is very little agreement of how this should be done. Here we present our current understanding of this problem and some partial solutions. We begin by characterizing evidence as information that is uncertain, incomplete, and sometimes inaccurate. Based on this characterization, we conclude that evidential reasoning requires both a method for pooling multiple bodies of evidence to arrive at a consensus opinion and some means of drawing the appropriate conclusions from that opinion. We contrast our approach, based on a relatively new mathematical theory of evidence, with those approaches bases on Bayesian probability models. We believe that our approach has some significant advantages, particularly its ability to represent and reason from bounded ignorance.
%0 Thesis %9 PhD Thesis %A Lowrance, John D. %T Dependency-Graph Models of Evidential Support %I Department of Computer and Information Science, University of Massachusetts, Amherst, MA %D 1982 %X Dependency-graph models of evidential support are formal systems capable of pooling and extending evidential information, while maintaining internal consistency. In this formalism, the likelihood of a proposition is represented as a subinterval of the until interval. The lower bound represents the degree of ``support'' provided a proposition by a body of evidence, and the upper bound represents the extent to which it remains ``plausible.'' The smaller this interval, the more precisely the probability of that proposition is known. Evidential information, extracted from the environment by (indivisible) sources of knowledge, enters these models in the form of probability
%0 Report %@ SRI Contra %A Garvey, Thomas D. and Lowrance, John D. %T Machine-Intelligence-Based Multisensor ESM System %C 333 Ravenswood Ave., Menlo Park, CA 94025 %I AI Center, SRI International %D 1981
%0 Conference Proceedings %A Garvey, Thomas D. and Lowrance, John D. and Fischler, Martin A. %T An Inference Technique for Integrating Knowledge from Disparate Sources %B Proceedings of the Seventh Joint Conference on Artificial Intelligence %C Menlo Park, CA %P 319-325 %D 1981 %X This paper introduces a formal method for integrating knowledge derived from a variety of sources for use in ``perceptual reasoning.’’ The formalism is based on the ``evidential propositional calculus.’’ a derivative of Shafer’s mathematical theory of evidence [Shafer 1976]. It is more general than either a Boolean or Bayesian approach, providing for Boolean and Bayesian inferencing when the appropriate information is available. In this formalism, the likelihood of a proposition A is represented as a subinterval, [s(A),p(A)], of the unit interval [0,1]. The evidential support for the proposition A is represented by s(A), while p(A) represents its degree of plausibility; p(A) can also be interpreted as the degree to which one fails to doubt A, p(A) being equal to one minus the evidential support for ~A. This paper describes how evidential information, furnished by a knowledge source in the form of a probability ``mass’’ distribution, can be converted to this interval representation; how, through a set of inference rules for computing intervals of dependent propositions, this information can be extrapolated from those propositions it directly bears upon, to those it indirectly bears upon; and how multiple bodies of evidential information can be pooled. A sample application of this approach, modeling the operation of a collection of sensors (a particular type of knowledge source), illustrates these techniques.
%0 Report %@ 79-6 %A Lowrance, John D. and Corkill, Daniel D. %T The Design of Grasper 1.0: A Programming Language Extension for Graph Processing %C Amherst, MA %I Department of Computer and Information Science, University of Massachusetts %D 1979 %X GRASPER 1.0 is a programming language extension. Once appended to a host language, GRASPER 1.0 introduces graphs, diagrams consisting of points connected by lines or arrows, as a primitive data type. The primary feature of GRASPER 1.0's design is that the language, its documentation, and its implementation all share a common organizational structure that groups GRASPER 1.0 primitives according to their scope of application and the underlying concepts from which they are formed. Although this report is of a descriptive nature, a similar approach might well be prescribed for other applications. GRASPER 1.0 is based on a small number of underlying concepts. GRASPER 1.0 primitives are constructed from these concepts according to a small set of rules. The name of each GRASPER 1.0 primitive systematically reflects its underlying concepts. This generative nature of the language organizes a large set of primitives in a cognitively efficient way. This makes GRASPER 1.0 easier to learn and retain; proves an indexing system for GRASPER 1.0 documentation; and serves as an outline for well-structured implementations. GRASPER 1.0 has been implemented with LISP 1.5 as the host language. This implementation supports a software-level virtual memory management system for graph storage. Spaces, user defined subgraphs, are used by the virtual memory manager to group logically related information on the same pages, helping to reduce paging. Multiple storage schemes allow users to optimize the way graphs are stored based on their particular applications.
%0 Report %@ 78-20 %A Lowrance, John D. %T Grasper 1.0 Reference Manual %C Amherst, MA %I Department of Computer and Information Science, University of Massachusetts %D 1978
%0 Journal Article %A Lowrance, John D. and Friedman, Daniel P. %T Hendrix's Model for Simultaneous Actions and Continuous Processes: An Introduction and Implementation %B International Journal of Man-Machine Studies %P 537-581 %V 9 %D 1977 %X This paper presents a self-contained introduction and implementation description to a simulation system for modeling simultaneous action and continuous processes (Hendrix, 1973). The essence of the system is described by a portion of its abstract: ``A new methodology for the construction of world models is presented. The central feature of this methodology is a mechanism which makes possible the modeling of (1) simultaneous, interactive processes, (2) processes characterized by a continuum of gradual change, (3) involuntarily activated processes (such as the growing of grass) and (4) times as a continuous phenomena.'' and by a recent review, Gains (1975): ``This is a fascination paper that will be of interest outside the ``artificial intelligence'' (AI) context in which it is written, from those concerned with simulating and controlling multi-element systems to those interested in operational definitions of concepts such as causality.'' Three robot world models are incrementally developed, each introducing a new modeling concept. World models, including a robot world (with sample output), electrical world, and a Turing world are also presented. The interactive operating environment represented permits the user to inspect and alter the run-time structure. A detailed account of the implementation is presented.
%0 Report %@ 77-1 %A Williams, Thomas D. and Lowrance, John D. %T Model-Building in the VISIONS High Level System %C Amherst, MA %I Department of Computer and Information Science, University of Massachusetts %D 1977 %X SYMBOLS, a general purpose model-building tool, has been designed and implemented in order to develop and test the semantic interpretation portion of the VISIONS system. Following the criterion of modularity, the data, processes and search concepts of model-building have been decomposed into units which are natural for the understanding of digitized outdoor scenes. Multiple-leveled structures are described for the representation of the model and processes. The strategy surrounding the control of processes in a huge search space is integrated into the system via a hierarchy of modular substrategies of control.
