mihai boicu gheorghe tecuci david schum disciple lta is a
play

Mihai Boicu, Gheorghe Tecuci, David Schum Disciple-LTA is a new type - PowerPoint PPT Presentation

Mihai Boicu, Gheorghe Tecuci, David Schum Disciple-LTA is a new type of personal cognitive assistant that: can rapidly acquire expertise in intelligence analysis, can train new intelligence analysts, and can help analysts analyze


  1. Mihai Boicu, Gheorghe Tecuci, David Schum

  2. Disciple-LTA is a new type of personal cognitive assistant that: ◦ can rapidly acquire expertise in intelligence analysis, ◦ can train new intelligence analysts, and ◦ can help analysts analyze complex hypotheses  through mixed-initiative reasoning,  allowing a synergistic integration of a human’s experience and creativity with an agent’s knowledge and speed, and  facilitating collaboration with complementary experts and their agents. ● Rapid and systematic hypothesis analysis ● Rapid acquisition of analytic expertise ● Assumption-based reasoning ● Learning to analyze based on source’s culture ● Evidence retrieval and representation ● Tutoring new analysts ● Automatic report generation ● Automatic test generation ● Collaboration and information sharing ● Easy integration with other tools and services

  3. Develop learning and problem solving agents that can be taught by subject matter experts to become knowledge-based assistants. The expert Disciple teaches Disciple continuously how to solve develops and refines D I S C I P L E problems in a way its knowledge base that resembles how to capture and the expert would better represent teach a student, 
 expert’s knowledge an apprentice or and problem solving a collaborator. strategies. There is no longer a distinction between knowledge base development and its maintenance.

  4. Assess whether It is unlikely that P 1 S 1 Al Qaeda has Al Qaeda has A complex hypothesis analysis 1. problem is successively reduced nuclear weapons. nuclear weapons. to simpler problems that either have known solutions or can be solved through evidence analysis. … P 11 S 11 P 1n S 1n Potentially relevant pieces of 2. evidence for the unsolved problems are identified. The pieces of evidence are 3. … analyzed to obtain solutions for S 111 P 111 P 11m S 11m the unsolved problems. The solutions of the simplest 4. problems are successively combined to obtain the solution … of the initial problem. S a P a P d S d 11m 11m 11m 11m National Intelligence Council’s standard estimative language

  5. TOC Reduction of a problem to its main sub-problems Abstract tree Detailed tree

  6. The ontology contains a hierarchical description of the domain concepts. 7

  7. The rules specify general reduction or synthesis steps To assess whether described with the there are states that may be willing concepts from the to sell nuclear ontology. weapons to an actor, one has to consider each ANALYSIS TREE nuclear state which is not an enemy of that P 1 S 1 actor and does not oppose the proliferation of nuclear weapons, and determine whether that state … P 11 S 11 P 1n S 1n may be willing to Partially sell nuclear weapons to that learned actor. rule … P 11m S 11m S 111 P 111

  8.  knowledge representation  user-agent communication  problem solving  knowledge acquisition and learning

  9. credibility credential Equivocal testimonial evidence veracity Testimonial evidence based upon direct observation objectivity has credibility credential EVD-Dawn-Mir01-02c observational sensitivity

  10. Wallflower’s Wallflower’s received Wallflower’s reported tes:mony about Wallflower’s translated tes:mony about Emir Z in English. recorded tes:mony about Emir Z in English. Wallflower’s tes:mony about Emir Z in English. tes:mony about Emir Z in Farsi. EVD‐ Emir Z in Farsi. EVD‐ Wallflower‐5 EVD‐ Wallflower‐4 EVD‐ Wallflower‐3 EVD‐ Wallflower‐2 Wallflower‐1 recording transla:on edi:ng transmission Wallflower • Competence • Credibility Clyde Bob Husam A. Marsha • Competence • Competence • Veracity Marsha • Competence Sony Recorder • Credibility • Credibility • Competence • Fidelity • Veracity • Reliability SN 247 • Fidelity • Reliability • Security

  11. A statement made by Osama bin Laden  represents his view  should not interfere with other (maybe contradictory) statements made by other sources, or by the analyst who taught Disciple.

  12. Osama bin Laden statement Osama bin Laden newspaper article terrorist has as source group elementary group testimonial piece of evidence evidence instance of instance of instance of is view of Al Qaeda Al Qaeda in-Dawn-Mir01-b Dawn-Mir01 instance of has as weapons states as view has as content uses as deterrent Dawn-Mir01-b nuclear has as author states as view weapons instance of Hamid Mir weapons 1 in-Dawn-Mir01-b

  13. Lesson Fragment: Hypothesis support from a piece of evidence Abstract reduction strategy Automatically generated illustration of the abstract strategy 17

  14. Veracity

  15. Sub-hypothesis: Assess whether other countries UDDI UDDI Hypothesis Hypothesis Hypothesis: Assess Check UDDI within the global community believe that Al Qaeda whether Al Qaeda has for registered Analysis Analysis has nuclear weapons. competence nuclear weapons. Broker Broker Perform Sub-hypothesis: Assess whether Al Qaeda makes 3 analysis credible claims to have nuclear weapons. 1 Ask broker for solutions 2 5 Receive solutions Solicit / receive solutions 6 4 Integrate Disciple-LTA Disciple-LTA Solutions Client Client Sub-hypothesis: Assess whether other 7 countries within the global community believe that Al Qaeda has nuclear weapons. Disciple-LTA Disciple-LTA Upload Solution: It is likely that other countries within reasoning Clients Clients tree in the global community believe that Al Qaeda Catalyst has nuclear weapons. Disciple LTA Disciple LTA Servers Servers Sub-hypothesis: Assess whether Al Qaeda makes Hypothesis Hypothesis credible claims to have nuclear weapons. Analysis Analysis Solution: It is almost certain that the Al Qaeda Web Service Web Service claims of having nuclear weapons are credible.

  16. 2. Disciple learns Learned Rule 1. The analyst extends reasoning rules the analysis logic To assess whether there are states that may be willing to sell nuclear weapons to an actor, one has one has to consider to consider Learned Rule each nuclear each nuclear state state and assess whether that state may be willing to sell nuclear weapons to that actor. 20

  17. 1. Disciple applies the learned rule 2. The analyst 3. Disciple-LTA refines critiques the the rule with an except- reasoning when condition Refined Rule This is wrong! France will not sell To assess whether there are nuclear states that may be willing to weapons to sell nuclear weapons to an actor, one one has to consider each has to consider each Al Qaeda nuclear state nuclear state and assess because it whether that state may be perceives it willing to sell nuclear weapons as an to that actor, except for except for the case in which the nuclear state enemy. is an enemy of that actor. 22

  18. Continuous Continuous direct direct adaptation of adaptation of test ev test ev the previously the previously learned rules learned rules to the to the evolution of evolution of the ontology. the ontology. IF <Problem> IF <Problem> PVS Condition PVS Condition Except-When Except-When PVS Condition PVS Condition THEN <Subproblem 1> THEN <Subproblem 1> … … <Subproblem m> <Subproblem m>

  19. Analytic Assistance Empowers the analysts through mixed-initiative reasoning for hypotheses analysis, collaboration with other analysts and experts, and sharing of information. Learning Tutoring Rapid acquisition Helps new and maintenance intelligence of subject matter analysts learn expertise in the reasoning intelligence processes analysis which involved in currently making takes years intelligence to establish, judgments is lost when and solving experts separate intelligence from service, and analysis is costly to problems. replace.

  20. Critical requirement for a new generation Disciple Learning and Tutoring Shell Agent Learning Theory of cognitive assistants Domain Knowledge Base Tutoring Knowledge Base Expertise modeling Multistrategy learning Mixed-initiative reasoning Reasoning with uncertainty STTR Phase II 25

Recommend


More recommend