Evaluating the Interpretability of the Knowledge Compilation Map: Communicating Logical Statements Effectively

Authors: Serena Booth, Christian Muise, Julie Shah

IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conduct a user study to evaluate whether knowledge compilation can aid logic interpretability. We find only sparse effects of knowledge compilation properties on interpretability. We discover some languages considered to be compilation-only are acceptable, while disjunctive normal form a representation assumed to be interpretable is not significantly more interpretable than other forms.
Researcher Affiliation Collaboration Serena Booth1 , Christian Muise2,3 and Julie Shah1 1MIT Computer Science and Artificial Intelligence Laboratory 2IBM Research 3MIT-IBM Watson AI Lab {serenabooth, julie a shah}@csail.mit.edu, christian.muise@ibm.com
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code Yes Our study procedures and source code are available at github.com/serenabooth/logic-interpretability.
Open Datasets No The paper describes generating data through simulations ('We simulate the agents and generate traces...') but does not provide concrete access information (link, DOI, repository, or formal citation to a public dataset) for this data or specify the use of a well-known public dataset.
Dataset Splits No The paper describes a user study and scenario questions but does not provide specific dataset split information (percentages, sample counts, or citations to predefined splits) for training, validation, or testing.
Hardware Specification No The paper mentions participants completed the study 'on-site' but does not provide specific hardware details (e.g., exact GPU/CPU models, memory amounts, or detailed computer specifications) used for running its experiments or simulations.
Software Dependencies No The paper mentions software tools like 'PMC PREPROCESSOR' and 'DSHARP' but does not provide specific version numbers for these or any other ancillary software components needed to replicate the experiment.
Experiment Setup No The paper describes the setup of a user study (e.g., number of questions, presentation format) but does not provide specific experimental setup details such as hyperparameter values, training configurations, or system-level settings for any computational models.