Admissibility in Probabilistic Argumentation

Authors: Nikolai Käfer, Christel Baier, Martin Diller, Clemens Dubslaff, Sarah Alice Gaggl, Holger Hermanns

JAIR 2022 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Enabled by the constraintbased approach, standard reasoning problems for probabilistic semantics can be tackled by SMT solvers, as we demonstrate by a proof-of-concept implementation. ... For example, the tool can compute a distribution maximizing the value of ct ... The tool and all experimental data are publicly available at https://www.perspicuous-computing.science/cpraa ... Details about our implementation and experimental studies on the vehicle example are provided in Section 6. We close the paper with concluding remarks and further work (Section 7).
Researcher Affiliation Academia Nikolai K afer EMAIL Christel Baier EMAIL Martin Diller EMAIL Clemens Dubslaff EMAIL Sarah Alice Gaggl EMAIL Faculty of Computer Science Technische Universit at Dresden Dresden, Germany Holger Hermanns EMAIL Saarland Informatics Campus Saarland University Saarbr ucken, Germany Institute of Intelligent Software Guangzhou, China
Pseudocode Yes Algorithm 1: Decide credulous acceptance for argument C under 1-Sat and θ = 1
Open Source Code Yes The tool and all experimental data are publicly available at https://www.perspicuous-computing.science/cpraa
Open Datasets No The paper uses a constructed
Dataset Splits No The paper introduces a conceptual
Hardware Specification Yes All experiments were conducted on an Intel i9-10900K machine with 64GB of RAM, running Ubuntu 20.10 and Python 3.8.6.
Software Dependencies Yes All experiments were conducted on an Intel i9-10900K machine with 64GB of RAM, running Ubuntu 20.10 and Python 3.8.6. ... First, SMT solvers like Z3 (de Moura & Bjørner, 2008) are able to handle arbitrary polynomial constraints in the existential theory of the reals. ... Second, linear-optimization solvers can be used for likelihood optimization and solution space enumeration, provided the selected semantics constraints are linear. ... available via CVXOPT (Andersen, Dahl, & Vandenberghe, 2014).
Experiment Setup Yes Additionally, we imposed the following constraints on the likelihoods of some of the arguments: µ(cl) = 0.7, µ(ld) = 0.7, µ(cr r) [0.7, 1], and µ(cr m) [0.4, 1]. ... A maximum 2% risk of a false positive detection was enforced via the conditional probability constraint µ(cr r cr m | cr) 0.98. In line with the sensor arrangement visualized in Figure 1, we enforced a scalar dependency of arguments cr r and cr m by the constraint µ(cr r) = 2 µ(cr m).