Rushing and Strolling among Answer Sets – Navigation Made Easy
Authors: Johannes Klaus Fichte, Sarah Alice Gaggl, Dominik Rusovac5651-5659
AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | To study the feasibility of our framework, we implemented the faceted answer set browser (fasb) on top of the clingo solver. In particular, we conducted experiments on three instance sets that range from large solution spaces to complex encodings in order to verify the following two hypotheses: |
| Researcher Affiliation | Academia | Johannes Klaus Fichte1, Sarah Alice Gaggl2, Dominik Rusovac2 1 Research Unit Database and Artificial Intelligence, TU Wien, Austria 2 Logic Programming and Argumentation Group, TU Dresden, Germany |
| Pseudocode | No | The paper contains formal definitions and mathematical notation but does not include any pseudocode or algorithm blocks. |
| Open Source Code | Yes | The implementation and experiments are publicly available (Fichte, Gaggl, and Rusovac 2021a,b). |
| Open Datasets | Yes | As input instance, we used the abstract argumentation framework A/3/ferry2.pfile-L3-C1-06.pddl.1.cnf.apx from the benchmark set of (ICCMA 17) (Gaggl et al. 2020). |
| Dataset Splits | No | The paper describes using specific problem instances and measuring performance during navigation steps, but it does not refer to or specify typical train/validation/test dataset splits as commonly found in machine learning experiments. |
| Hardware Specification | Yes | experiments were run on an eight core Intel i7-10510U CPU 1.8 GHz with 16 GB of RAM, running Manjaro Linux 21.1.1 (kernel 5.10.59-1-MANJARO). |
| Software Dependencies | No | The paper states that the implementation uses 'clingo solver' but does not specify a version number for clingo or any other software dependencies with version numbers. |
| Experiment Setup | Yes | Thus, we run three iterations of random navigation steps in each of the implemented modes, to simulate a user and avoid bias regarding the choice of steps. For go, sgo-fc, and sgo-abs, we use the --random-safe-walk call, which in the provided mode performs random steps until the current route is maximal safe, e.g., in sgo-fc and sgoabs it computes maximal weighted facets and then chooses one of them to activate randomly. Since, in practice, using expl-fc and expl-abs, we do not necessarily aim to arrive at a unique solution, we use --random-safe-steps for expl-fc and expl-abs and provide the maximum number n of steps among iterations in go, which performs n random steps in the provided mode. |