September 30, 2019, at CP 2019, Stamford, Connecticut, USA
Note: For the Papers and the Holy Grail Challenge, both papers and slides are available; for the Invited Talks the slides are available.
9:00 – 9:10: Welcome. Gene Freuder
9:10 – 10:30: Papers
- Abstract_Model_Generation_in_Interactive_Consultant. Slides. Pierre Carbonnelle, Gerda Janssens, and Marc Denecker
- Learning the Parameters of Global Constraints for Medical Scheduling. Slides. Emilie Picard-Cantin, Mathieu Bouchard, Claude-Guy Quimper, and Jason Sweeney
- Constraint Acquisition Via Classification. Slides. S. D. Prestwich, E. C. Freuder D. Brown, B. O’Sullivan
- Learning Scheduling Models from Event Data. Slides. Arik Senderovich, Kyle E. C. Booth, J. Christopher Beck
10:30 – 11:00: Coffee Break
11:00 – 12:30: Invited Talks
- A Cognitive Modeling Assistant to Optimize Complex Decisions. Xavier Ceugniet. IBM Analytics
- Empirical Model Learning. Michele Lombardi. University of Bologna
- In Pursuit of the Holy Grail of Natural Language Understanding: Past, Present, and Future. Daniel Khashabi. Allen Institute for Artificial Intelligence
12:30 – 2:00 Lunch
2:00 – 3:30: Holy Grail Challenge
- 2:00 – 2:30: Solving Logic Grid Puzzles with an Algorithm that Imitates Human Behavior. Slides. Guillaume Escamocher and Barry O’Sullivan
- 2:30 – 3:00: *Challenge Winner* User-Oriented Solving and Explaining of Natural Language Logic Grid Puzzles. Slides. Jens Claes, Bart Bogaerts, Rocsildes Canoy, Tias Guns
- 3:00 – 3:30: Discussion, Voting, Planning the next Challenge
3:30 – 4:00: Coffee Break
4:00 – 5:30: Interactive
- Brainstorming: The path to the Holy Grail
- Planning: Building a community
- Networking: Working together
Over twenty years ago the paper “In Pursuit of the Holy Grail” proposed that Constraint Programming was well-positioned to pursue the Holy Grail of computer science: the user simply states the problem and the computer solves it. This workshop will look at progress towards that goal, in particular regarding work on automating:
- Problem Acquisition: learning, debugging, maintaining, etc.
- Model Reformulation: transformation for efficient solution, redundant models, etc.
- Solver Construction: adaptive parameter tuning, automated selection from portfolios, etc.
- User Explanation: reasons for failure, implications for choices, etc.
Of special interest is the intersection of the Holy Grail goal with the increasing attention being paid to machine learning, explainable AI, Human-Aware AI, Human-AI Collaboration, and intelligent software assistants.
Chair: Eugene Freuder, University College Cork, Ireland: email@example.com
Luc De Raedt, Katholieke Universiteit Leuven, Belgium
Lars Kotthoff, University of Wyoming, USA
Ian Miguel, University of St Andrews, Scotland
Submissions may be of any length, and in any format. They may be abstracts, position papers, technical papers, or demos. They may review your own previous work or survey a topic area. They may present new research or suggest directions for further progress. They may propose research roadmaps, demonstration domains, or collaborative projects. They may be proposals for measuring progress, and, in particular, for data sets or competitions to stimulate and compare progress.
Authors may make multiple submissions if they wish. All submissions that appropriately address the topic of the workshop will be accepted as is, without further revision, and will be made available at the workshop website.
At least one author of every accepted submission must attend the workshop and pay the workshop fee; otherwise any presentation (and submission) will be withdrawn from the proceedings (if any) and program.
Submissions should be emailed, by August 1, 2019, in PDF form, with subject line “PTHG-19 Submission”, directly to the Workshop chair, at: firstname.lastname@example.org.
Decisions on submissions will be emailed at least 1 week before the early registration date for the conference.
The 2019 Holy Grail Challenge
The 2019 Progress Towards the Holy Grail Workshop at CP-19 invites entrants for the 2019 Holy Grail Challenge. The Challenge objective is to provide automated processing of a simple, restricted class of problems, logic puzzles, specifically what are sometimes called logic grid puzzles.
Given the short time frame, entries will be accepted in two categories:
- Proposals: Short papers describing methods for implementing all or part of the Challenge
- Results: Short papers describing implementations of all or part of the Challenge
The logicgridpuzzles.com website provides puzzles of varying difficulty. For each puzzle it provides a description and clues in English language, and a labelled grid.
The Challenge: For puzzles at the website:
- Accept as input the description and clues in English. If needed you can also use the information contained in the grid labels; to avoid the need for machine vision you can convert this by hand into a suitable input format. If needed, you may also provide in a suitable input format the general rule of logic grid puzzles that each option is only used once; for example, when the Dragon Slayer puzzle asks “What are the banner colors of each knight?” it is understood that each knight has a different color.
- Transform the input into an appropriate CSP model.
- Solve the CSP appropriately.
- Provide an appropriate English language explanation of how the solution was obtained.
Specifying the meaning of “appropriate” is part of the Challenge. For example, one model or solution method might be appropriate for swift solution, while a different one might be appropriate for obtaining people-friendly explanations, one model or solution method might be appropriate for difficult problems while a different one might be appropriate for easy problems.
Entries may be:
- Complete: Addressing all 4 steps of the challenge
- Partial: Addressing either the acquisition or explanation side of the problem alone, i.e. by omitting step 4, or by providing an initial CSP model by hand.
Entrants are free to use off-the-shelf tools, e.g. a natural language toolkit, or a constraint modeling language. Implementations do not need to interface directly with the puzzle website.
Submissions to the Challenge:
- Entries may be submitted any time up to September 1, 2019. Any serious entry will be accepted, and entrants notified within a week. (For notification by the early registration date for the Conference, submit by August 8, 2019.)
- Email entries, in PDF form, with subject line “HGC Entry”, directly to the Workshop chair, at: email@example.com.
- There are no formatting requirements.
- The entries will be posted at the Workshop website, and the entrants will present their entry to the Workshop audience, which will vote to determine award-winning entries.
- At least one author of every accepted entry must attend the workshop and pay the workshop fee; otherwise any presentation (and entry) will be withdrawn from the proceedings (if any) and program.
The following may help entrants come to grips with the challenge: