Towards Semi-Automatic Learning-Based Model Transformation

Zeighami, Kiana; Leo, Kevin; Tack, Guido; Banda, Maria García de la

doi:10.1007/978-3-319-98334-9_27

Cited by 4 publications

(6 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The initial stages of our method (all substeps in step 1: finding core candidates) are automatic, thanks to the use of core-guided optimisation to generate the cores and their information, and the use of existing MINIZINC infrastructure to collect and rename the cores, and identify the core candidates. While substep 2.1 (finding patterns among the cores) is currently done manually, it can be automated using similar technology to that used by [24] to identify patterns among nogoods.…”

Section: Automating the Methodsmentioning

confidence: 99%

“…dzn instance: Typically, solver writers who want to interpret such names, must examine the compiled instance output to see what these variables might refer to. In [24], the authors used a source map produced by the MINIZINC compiler, to map instance variables back to variables and expressions in the original model. Herein, we use the same method to link back the core variables, which for the above core results in: The variables can now be easily recognised (from the objective) as the earliness of task 16, and the tardiness of task 25.…”

Section: Step 1: Finding Core Candidatesmentioning

confidence: 99%

“…Note that we always ignore the literals' bound (e.g., 4 and 1 in the above core). While we currently find these patterns manually, the method described in [24] for finding nogood patterns across instances, can be easily adapted to cores.…”

Section: Step 2: Selecting Good Candidatesmentioning

confidence: 99%

“…This allows LCG solvers to infer nogoods (i.e., reasons for failure), and use them to avoid repeatedly exploring infeasible subproblems. Previous work [24] showed that the nogoods inferred by LCG solvers for a model's instances can be used to identify (a) existing constraints that may Partly funded by Australian Research Council grant DP180100151. be strengthened, and (b) new redundant constraints on the existing model variables that are likely to increase performance.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Core-Guided Model Reformulation

Leo

Gange

Banda

et al. 2020

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Constraint propagation and SAT solvers often underperform when dealing with optimisation problems that have an additive (or separable) objective function. The core-guided search introduced by MaxSAT solvers can overcome this weakness by detecting and exploiting cores: subsets of the objective components that cannot collectively take their lower bounds. This paper shows how to use the information collected during core-guided search, to reformulate the objective function for an entire class of problems (those captured by the problem model). The resulting (currently manual) method is examined on several case studies, with very promising results.

show abstract

Section: Automating the Methodsmentioning

confidence: 99%

Section: Step 1: Finding Core Candidatesmentioning

confidence: 99%

Section: Step 2: Selecting Good Candidatesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Core-Guided Model Reformulation

Leo

Gange

Banda

et al. 2020

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…The QuickXplain method [8] for example uses a dichotomic approach that recursively partitions the constraints to find a minimal conflict set. Many other papers consider the same goal and search for explanations of over-constrainedness [17,18]. A minimal set of conflicting constraints is often called a minimal unsatisfiable subset (MUS) or minimal unsatisfiable core [19].…”

Section: Related Workmentioning

confidence: 99%

A framework for step-wise explaining how to solve constraint satisfaction problems

Bogaerts,

Gamba,

Guns

2020

Preprint

View full text Add to dashboard Cite

We explore the problem of step-wise explaining how to solve constraint satisfaction problems, with a use case on logic grid puzzles. More specifically, we study the problem of explaining the inference steps that one can take during propagation, in a way that is easy to interpret for a person. Thereby, we aim to give the constraint solver explainable agency, which can help in building trust in the solver by being able to understand and even learn from the explanations. The main challenge is that of finding a sequence of simple explanations, where each explanation should aim to be as cognitively easy as possible for a human to verify and understand. This contrasts with the arbitrary combination of facts and constraints that the solver may use when propagating. We propose the use of a cost function to quantify how simple an individual explanation of an inference step is, and identify the explanation-production problem of finding the best sequence of explanations of a CSP. Our approach is agnostic of the underlying constraint propagation mechanisms, and can provide explanations even for inference steps resulting from combinations of constraints. In case multiple constraints are involved, we also develop a mechanism that allows to break the most difficult steps up and thus gives the user the ability to zoom in on specific parts of the explanation. Our proposed algorithm iteratively constructs the explanation sequence by using an optimistic estimate of the cost function to guide the search for the best explanation at each step. Our experiments on logic grid puzzles show the feasibility of the approach in terms of the quality of the individual explanations and the resulting explanation sequences obtained.

show abstract