Plan Explanations as Model Reconciliation: Moving Beyond Explanation as Soliloquy

Authors: Tathagata Chakraborti, Sarath Sreedharan, Yu Zhang, Subbarao Kambhampati

IJCAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we present a preliminary evaluation of the efficiency of our algorithms for generating explanations in randomly generated problems in a few benchmark planning domains.
Researcher Affiliation Academia Tathagata Chakraborti and Sarath Sreedharan and Yu Zhang and Subbarao Kambhampati School of Computing, Informatics, and Decision Systems Engineering Arizona State University, Tempe, AZ 85281 USA { tchakra2, ssreedh3, yzhan442, rao } @ asu.edu
Pseudocode Yes Algorithm 1 Search for Minimally Complete Explanations; Algorithm 2 Search for Minimally Monotonic Explanations
Open Source Code Yes The latest version of the code will be available at https://goo.gl/Bybq7E.
Open Datasets No The paper states: "We use three planning domains Blocks World, Logistics and Rover for our experiments." While these are well-known planning domains, the paper does not provide specific citations, links, or access information for publicly available datasets, nor does it specify how any custom-generated data for these domains can be accessed or recreated.
Dataset Splits No The paper does not provide specific dataset split information for training, validation, or testing, such as percentages, sample counts, or references to predefined splits.
Hardware Specification Yes The results reported here are from experiments run on a 12 core Intel(R) Xeon(R) CPU with an E5-2643 v3@3.40GHz processor and a 64G RAM.
Software Dependencies No The paper mentions software like "Fast-Downward", "VAL", and "pyperplan" along with citations. However, it does not provide specific version numbers for these components, which is required for reproducibility.
Experiment Setup No The paper describes how the human models were created (e.g., "randomly removing parts (preconditions and effects) of the action model") and the domains used. However, it does not provide specific experimental setup details such as hyperparameters for the search algorithms (e.g., A* parameters, heuristics other than admissibility) or system-level training settings.