Carmel Domshlak

Associate Professor
Technion

   Faculty of Industrial Engineering & Management
(IE&M)

tel: +972.4.8294437 | fax: +972.4.8295688



Links :  Research Group  |  Publications  |  Tutorials & Overviews  |  Software  |  World Exploration


Education : 1996 - 2002   Ph.D. and M.Sc. at Ben-Gurion University (cs@bgu), Beer-Sheva, Israel   [thesis]
1990 - 1993   B.Sc. at Ben-Gurion University (cs@bgu), Beer-Sheva, Israel
 
Curriculum Vitae : pdf   (kind of updated ...)


Research & Research Group


My primary research interest is in Artificial Intelligence (AI) computational and modeling problems, with an emphasis on interesting structure/complexity connections that can be found in different tasks of automated reasoning such as domain-independent planning, multi-agent system design, and user-centric information systems (via preferential reasoning). The recent focus of my work has been on heuristic functions for domain-indepent forward-search planning, and on parametric complexity of various fragments of planning tasks. Tangentially, I am also interested in problems that lie on the boundary between computer science and social choice, e.g., various settins of rank aggregation, voting, recommendation, etc. (Warning: my "research focus" appears to be quite a moving target as there are simply too many interesting questions on AI's research agenda ...)

Research Area Research Style
                
     Automated Planning
     Information Retrieval & Ranking
     Preferential Reasoning
                
     Formal Constructs & Analysis
     Implementation & Empirical Analysis
     Multi-Disciplinary Connections

GROUP MEMBERS

    
   Yagil Engel       Postdoc   
 
   Michael Katz       PhD       Structural Patterns Heuristics   
   Erez Karpas       PhD       Learning for Planning   
     (co-advised by Shaul Markovitch)   
 
   Sagi Lepler       MSc       Optimized Abstraction Heuristics   
   Kolman Vornovitsky       MSc       Bottom-Up Abstractions for Decision-Theoretic Planning   
   Mattan Winaver       MSc       Automatic Query Expansion Selection   
     (co-advised by Oren Kurland)   

ALUMNI

    
   Alex Naidich       MSc       Collaboration Reasoning about User Preferences   
     (effectively co-advised by Thorsten Joachims)       2008 ↦ Intel R&D Center, Haifa, Israel
   Evgeny Televitckiy       MSc       Opinion Aggregation for Double-Cold Start in Recommender Systems       2008 ↦ Airspan, Israel
   Victor Dweck       MSc       Non-directional Search for Factored Planning       2008 ↦ Kadima Asset Management, Rio de Janeiro, Brazil
   Vitaly Mirkis       MSc       STRIPS Planning with No Negative Effects       2007 ↦ Intel R&D Center, Haifa, Israel
   Naama Zwerdling       MSc       Effective ranking methods for desktop search       2007 ↦ IBM Research Labs, Haifa, Israel
       (co-advised by Sara Cohen)   
   Ruslana Makovetsky       MSc       Adopting Machine Learning Approaches to the Novelty Detection Task       2007 ↦ Microsoft, Haifa, Israel


(SOME) TOPICS FOR MSC/PHD RESEARCH

    
    
 
    Structural patterns beyond fork decomposition   
  
    Extending the palette of poly-time fragments of cost-optimal planning that can be used in structural-pattern heuristics beyond the fork-decomposition framework suggested in ICAPS'08.
 
   Abstraction-based heuristics for more complicated (than classical) planning formalisms   
  
    This is quite an open direction so things will have to start with a formal feasibility study.
 
   Mixed strategic/tactic planning for physical autonomous systems (UAVs)   
   
    Combining efficiently and effectively AI (strategic) planning with motion (tactic) planning is both interesting and challenging. This particluar project (under the guidance of Tal Shima and myself) considers this "mixture planning" problem for groups of autonomous aerial vehicles (UAVs).
 
   Empowering the pruning power of the A* search for domain-independent classical planning   
  
    Interesting (and super-challenging) project that aims at combining (and developing) graph-theoretic concepts around the notion of "graph automorhism", and exploiting it to empower the A* search. For why it is so important, see a recent paper by Malte Helmert and Gabi Röger How good is almost perfect?
 
   Topics in Information Meta-Retrieval   
   
    There are numerous ideas on the boundary of Information Retrieval and Social Choice that Oren Kurland and myself are currently exploring. For further details, please contact directly one of us.
 
   Building benchmarks for reasoning about preferences   
   
    Numerous knowledge-representation and reasoning tools have been developed over the years for handling people's preference statements (see, e.g., our tutorial). However, the comparison between them is quite problematic due to the absence of evaluation benchmarks. Can we provide the community with such benchmarks?



PUBLICATIONS



   [Recent stuff (2009)]


   [Planning as Heuristic Search]

  • Landmarks, Critical Paths and Abstractions: What's the Difference Anyway? [pdf, slides]
    Malte Helmert, Carmel Domshlak
    ICAPS-09. 19th International Conference on Automated Planning and Scheduling, Thessaloniki, Greece, September 2009.
    Best Paper Award.

  • Structural-Pattern Databases [pdf, slides]
    Michael Katz, Carmel Domshlak
    ICAPS-09. 19th International Conference on Automated Planning and Scheduling, Thessaloniki, Greece, September 2009.

  • Cost-optimal Planning with Landmarks [pdf, slides]
    Erez Karpas, Carmel Domshlak
    IJCAI-09. 21st International Joint Conference on Artificial Intelligence, Pasadena, CA, USA, July 2009.

  • Structural Pattern Heuristics via Fork Decomposition [pdf, technical report, slides]
    Michael Katz, Carmel Domshlak
    ICAPS-08. 18th International Conference on Automated Planning and Scheduling, Sydney, Australia, September 2008.

  • Optimal Additive Composition of Abstraction-based Admissible Heuristics [pdf,technical report, slides]
    Michael Katz, Carmel Domshlak
    ICAPS-08. 18th International Conference on Automated Planning and Scheduling, Sydney, Australia, September 2008.

  • Cost-Sharing Approximations for h+ [pdf, slides]
    Vitaly Mirkis, Carmel Domshlak
    ICAPS-07. 17th International Conference on Automated Planning and Scheduling, pp 240-247, Providence, RI, September 2007.

  • Probabilistic Planning via Heuristic Forward Search and Weighted Model Counting [abstract, pdf]
    Carmel Domshlak, Joerg Hoffmann
    Journal of Artificial Intelligence Research, volume 30, pp 565-620, 2007.

    • Fast Probabilistic Planning Through Weighted Model Counting [pdf, slides]
      Carmel Domshlak, Joerg Hoffmann
      ICAPS-06. 16th International Conference on Automated Planning and Scheduling, pp 243-252, The English Lake District, U.K., September 2006.

  • Multi-agent opportunistic planning and plan execution [pdf]
    Carmel Domshlak, James Lawton
    ICTAI-04. 16th IEEE International Conference on Tools with Artificial Intelligence, Boca Raton, Florida, November 2004.


   [Planning: Structure, Complexity and Tractability]

  • Planning Games [pdf, slides]
    Ronen Brafman, Carmel Domshlak, Yagil Engel, and Moshe Tennenholtz
    IJCAI-09. 21st International Joint Conference on Artificial Intelligence, Pasadena, CA, USA, July 2009.

  • From One to Many: Planning for Loosely Coupled Multi-Agent Systems [pdf, slides]
    Ronen I. Brafman, Carmel Domshlak
    ICAPS-08. 18th International Conference on Automated Planning and Scheduling, Sydney, Australia, September 2008.
    Best Paper Award.

  • Structural Pattern Heuristics via Fork Decomposition [pdf, slides]
    Michael Katz, Carmel Domshlak
    ICAPS-08. 18th International Conference on Automated Planning and Scheduling, Sydney, Australia, September 2008.
    Comment: the focus of this paper is on domain-independent planning heuristics, but it introduces three new tractable fragments of cost-optimal planning (that are shown to be not of a theoretical interest only).

  • New Islands of Tractability of Cost-Optimal Planning [abstract, pdf]
    Michael Katz, Carmel Domshlak
    Journal of Artificial Intelligence Research, volume 32, pp 203-288, 2008.

    • Structural Patterns of Tractable Sequentially-Optimal Planning [pdf, slides]
      Michael Katz, Carmel Domshlak
      ICAPS-07. 17th International Conference on Automated Planning and Scheduling, pp 200-207, Providence, RI, September 2007.

  • Factored planning: How, When, and When Not [pdf, slides]
    Ronen I. Brafman, Carmel Domshlak
    AAAI-06. 21st National Conference on Artificial Intelligence, Boston, MS, August 2006.

  • Friends or Foes? An AI Planning Perspective on Abstraction and Search [pdf]
    Joerg Hoffmann, Ashish Sabharwal, Carmel Domshlak
    ICAPS-06. 16th International Conference on Automated Planning and Scheduling, pp 294-303, The English Lake District, UK, June 2006.
    Nominated for the Best Paper Award.

  • Sensor Networks and Distributed CSP: Communication, Computation and Complexity [abstract & pdf]
    Ramon Béjar, Carmel Domshlak, Cezar Fernández, Carla Gomes, Bhaskar Krishnamachari, Bart Selman, Magda Valls
    Artificial Intelligence, Special Issue on Distributed Constraint Satiisfaction, volume 161, number 1-2, pp 117-147, 2005.

    • Grid-based Sensor DCSP [ps]
      Ramon Béjar, Carmel Domshlak, Cezar Fernández, Carla Gomes, Bart Selman, Magda Valls
      IJCAI-03. 18th International Joint Conference on Artificial Intelligence, Acapulco, Mexico, August 2003.
      Short paper.

  • Structure and Complexity in Planning with Unary Operators [abstract, pdf]
    Ronen I. Brafman, Carmel Domshlak
    Journal of Artificial Intelligence Research, volume 18, pp 315-349, 2003.

    • Structure and Complexity in Planning with Unary Operators [ps]
      Carmel Domshlak, Ronen I. Brafman
      AIPS-02 (ICAPS-02). 6th International Conference on Artificial Intelligence Planning and Scheduling, pp 34-43, Toulouse, France, April 2002.

  • Multi-Agent Offline Coordination: Structure and Complexity [ps]
    Carmel Domshlak, Yefim Dinitz
    ECP-01 (ICAPS-01). 6th European Conference on Planning, pp 34-43, Toledo, Spain, July 2002.


   [Planning and Cognition]

  • Distinguishing Between Intentional and Unintentional Sequences of Actions
    Elisheva Bonchek Dokow, Gal A. Kaminka, and Carmel Domshlak
    ICCM-09, International Conference on Cognitive Modeling 2009.


   [Preferential Reasoning: Parametrization-Light Techniques]

  • Efficient and Non-Parametric Reasoning over User Preferences [abstract & pdf]
    Carmel Domshlak, Thorsten Joachims
    User Modeling and User-Adapted Interaction journal, Special Issue on Statistical and Probabilistic Methods for User Modeling, volume 17, number 1-2, pp 41-69, 2007.

    • Unstructuring User Preferences: Efficient Non-Parametric Utility Revelation [pdf, slides]
      Carmel Domshlak, Thorsten Joachims
      UAI-05. 21st Conference on Uncertainty in Artificial Intelligence, pp 169-177, Boston, MS, August 2006.


   [Preferential Reasoning: Graphical Models]

  • Graphically Structured Value-Function Compilation [abstract & pdf]
    Ronen I. Brafman, Carmel Domshlak
    Artificial Intelligence journal, volume 172, number 2-3, pp 325-349, 2008.

    • Compact Value-Function Representations for Qualitative Preferences [pdf, slides]
      Ronen I. Brafman, Carmel Domshlak, Tanya Kogan
      UAI-04. 20st Conference on Uncertainty in Artificial Intelligence, pp 51-58, Banff, Canada, August 2006.

  • Preferences over Sets [pdf, slides]
    Ronen I. Brafman, Carmel Domshlak, Solomon E. Shimony, Yael Silver
    AAAI-06. 21st National Conference on Artificial Intelligence, Boston, MS, August 2006.

  • Hard and Soft Constraints for Reasoning about Qualitative Conditional Preferences [pdf]
    Carmel Domshlak, Steve Prestwich, Fancesca Rossi, Brent Venable, Toby Walsh
    Heuristics journal, Special Issue on Preferences, volume 12, number 4-5, pp 263-285, 2006.

    • Reasoning about Soft Constraints and Conditional Preferences [pdf]
      Carmel Domshlak, Francesca Rossi, Brent Venable, Toby Walsh
      IJCAI-03. 18th International Joint Conference on Artificial Intelligence, pp 215-220, Acapulco, Mexico, August 2003.

  • On Graphical Modeling of Preference and Importance [abstract, pdf]
    Ronen I. Brafman, Carmel Domshlak, Solomon E. Shimony
    Journal of Artificial Intelligence Research, volume 25, pp 389-424, 2006.

    • Introducing Variable Importance Tradeoffs into CP-nets [ps]
      Ronen I. Brafman, Carmel Domshlak
      UAI-02. 18th Conference on Uncertainty in Artificial Intelligence, pp 69-76, Edmonton, Canada, August 2002.

  • Qualitative Decision Making in Adaptive Presentation of Structured Information [abstract, pdf, advanced draft (ps)]
    Ronen I. Brafman, Carmel Domshlak, Solomon E. Shimony
    ACM Transactions on Information Systems, volume 22, number 4, pp 503-539, 2004.

    • Preference-based Configuration of Web Page Content [ps]
      Carmel Domshlak, Ronen I. Brafman, Solomon E. Shimony
      IJCAI-01. 17th International Joint Conference on Artificial Intelligence, pp 1451-1456, Seattle, WS, August 2001.

  • CP-nets: A Tool for Representing and Reasoning with Conditional Ceteris Paribus Preference Statements [abstract, pdf]
    Craig Boutilier, Ronen I. Brafman, Carmel Domshlak, Holger Hoos, David Poole
    Journal of Artificial Intelligence Research, volume 21, pp 135-191, 2004.
    2005 IJCAII-JAIR Best Paper Prize.

  • Preference-based Constraint Optimization with CP-nets [advanced draft (ps)]
    Craig Boutilier, Ronen I. Brafman, Carmel Domshlak, Holger Hoos, David Poole
    Computational Intelligence, Special Issue on Preferences, volume 20, number 2, pp 137-157, 2004.

  • CP-nets - reasoning and consistency testing [ps]
    Carmel Domshlak, Ronen I. Brafman
    KR-02. 18th International Conference on Principles of Knowledge Representation and Reasoning , pp 121-132, Toulouse, France, August 2002.
    Comments: Some stuff from this paper can be read in more detail in JAIR's "big CP-nets paper", but for a detailed discussion of the rest of the results (on cyclic CP-nets) I would refer the interested readed to my thesis.

  • On Recursively Directed Hypercubes [abstract, pdf]
    Carmel Domshlak
    Journal of Combinatorics, volume 9, number 1, R23, 2002.


   [Information Retrieval and Ranking]

  • Navigating in the Dark: Modeling Uncertainty in Ad Hoc Retrieval using Multiple Relevance Models [pdf]
    Nataly Soskin, Oren Kurland, Carmel Domshlak
    ICTIR-09. 2nd International Conference on the Theory of Information Retrieval, Cambridge, UK, September 2009.

  • A Rank-Aggregation Approach to Searching for Optimal Query-Specific Clusters [pdf]
    Oren Kurland, Carmel Domshlak
    SIGIR-08. 31st International Conference on Research and Development in Information Retrieval, Singapore, July 2008.

  • On Ranking Techniques for Desktop Search [abstract & pdf]
    Sara Cohen, Carmel Domshlak, Naama Zwerdling
    ACM Transactions on Information Systems, volume 26, number 2, P11, 2008.

    • On Ranking Techniques for Desktop Search [pdf]
      Sara Cohen, Carmel Domshlak, Naama Zwerdling
      WWW-07. 16th International Conference on World Wide Web, Banff, Canada, July 2007.
      Short paper.

  • Rank Aggregation for Schema Matching [pdf]
    Carmel Domshlak, Avigdor Gal, Haggai Roitman
    IEEE Transactions on Knowledge and Data Engineering, volume 19, number 4, pp 538-55, 2007.

  • Towards Robust Query Expansion: Model Selection in the Language Modeling Framework [pdf]
    Mattan Winaver, Oren Kurland, Carmel Domshlak
    SIGIR-07. 30th International Conference on Research and Development in Information Retrieval, Amsterdam, Holland, July 2007.
    Short paper.

  • Better than the Real Thing? Iterative Pseudo-Query Processing using Cluster-based Language Models [pdf]
    Oren Kurland, Lillian Lee, Carmel Domshlak
    SIGIR-05. 28th International Conference on Research and Development in Information Retrieval, Salvador, Brazil, July 2005.


   [Probabilistic Reasoning]

  • On Probabilistic Reasoning in Bayes Nets with Mutual Exclusion and Conditional Independence [pdf (including the proofs appendix)]
    Solomon E. Shimony, Carmel Domshlak
    International Journal of Intelligent Systems, volume 19, number 8, pp 703-726, 2004.

    • Efficient Probabilistic Reasoning in Bayes Nets with Mutual Exclusion and Context Specific Independence [ps]
      Carmel Domshlak, Solomon E. Shimony
      FLAIRS-03. 16th International FLAIRS Conference, St. Augustine, FL, 2003.

  • Complexity of Probabilistic Reasoning in Directed-Path Singly Connected Bayes Networks [ps]
    Solomon E. Shimony, Carmel Domshlak
    Artificial Intelligence journal, volume 151, number 1-2, pp 213-225, 2003.

  • FlexiMine - A Flexible Platform for KDD Research and Application Development [pdf]
    Rachel Ben-Eliyahu-Zohary, Carmel Domshlak, Ehud Gudes, Natalia Liusternik, Amnon Meisels, Tzachi Rosen, Solomon E. Shimony
    Annal of Mathematics and Artificial Intelligence, volume 39, number 1-2, pp 175-204, 2003.
    Preliminary version in KDD-98, 4th International Conference on Knowledge Discovery in Databases, pp 184-188, New-York, NY, 1998.

  • Cost-sharing in Bayesian Knowledge Bases [ps]
    Solomon E. Shimony, Carmel Domshlak, Eugene Santos
    UAI-97. 13th Conference on Uncertainty in Artificial Intelligence, pp 421-428, Providence, RI, August 1997.




Tutorials & Overviews


TUTORIALS
  • Representing, Eliciting, and Reasoning with Preferences [abstract & slides]
    Ronen I. Brafman, Carmel Domshlak
    AAAI-07. 22nd Conference on Artificial Intelligence, Tutorial Forum, Vancouver, British Columbia, July 2007.
    KI-07. 30th German Conference on Artificial Intelligence, Tutorial Forum, Osnabruck, Germany, September 2007.

  • Graphical Modeling of Qualitative Preferences [slides]
    Carmel Domshlak
    Dagstuhl. Seminar on Preferences: Specification, Inference, Applications (04271), Tutorial, 2004.

  • Representing, Eliciting, and Reasoning with Preferences [ppt]
    Ronen I. Brafman, Carmel Domshlak
    CP-03. 9th International Conference on Principles and Practice of Constraint Programming, Tutorial Forum, September, 2003.




Software

  • Probabilistic-FF   [Domshlak & Hoffmann, ICAPS-06, JAIR-08] is a domain independent planning system developed in collaboration with Jörg Hoffmann (and implemented solely by Master Jörg!). The system deals with conformant probabilistic planning problems in which actions have stochastic effects, and the initial configuration of the agent is given by a (Bayesian net) probability distribution over the world states.

  • HSP with set-based, cost-sharing heuristics  [Mirkis & Domshlak, ICAPS-07]
    (soon ...)

  • FactPlan  is a domain independent planning system developed in collaboration with Victor Dweck and Ronen I. Brafman. The systems deals with locally-optimal planning, trying to make practical sense of the ideas in [Brafman & Domshlak, AAAI-06].
    (soon ...)



World Exploration


visited 30 states (13.3%)
Create your own visited map of The World