Why Can't You Convince Me? Modeling Weaknesses in Unpersuasive Arguments
Authors: Isaac Persing, Vincent Ng
IJCAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | we (1) annotate a corpus of debate comments with not only their persuasiveness scores but also the errors they contain, (2) propose an approach to persuasiveness scoring and error identification that outperformscompeting baselines, and (3) show that the persuasiveness scores computed by our approach can indeed be explained by the errors it identifies. |
| Researcher Affiliation | Academia | Isaac Persing and Vincent Ng Human Language Technology Research Institute University of Texas at Dallas Richardson, TX 75083-0688 {persingq,vince}@hlt.utdallas.edu |
| Pseudocode | No | The paper describes its approach and features in text, but does not include any structured pseudocode or algorithm blocks. |
| Open Source Code | No | Finally, we make our annotated dataset publicly available.1 Given the difficulty of obtaining annotated data for this task, we believe our dataset will be a valuable resource to the NLP community. Footnote 1: See http://www.hlt.utdallas.edu/ persingq/ Debate/ for a complete list of our annotations. |
| Open Datasets | Yes | We use as our corpus a randomly selected subset of 165 debates obtained from the International Debate Education Association (IDEA) website2. These debates cover a wide range of topics including politics, economics, religion, and science. Footnote 2: http://idebate.org/ |
| Dataset Splits | Yes | Five-fold cross-validation results of the six baselines and OUR approach on the AP scoring task and the five error severity tasks as measured by E, ME, and PC are shown in the three subtables of Table 6.8 To ensure generalizability across new topics, we distribute arguments into folds based on the motions they respond to. |
| Hardware Specification | No | The paper does not provide specific details regarding the hardware used for running experiments. |
| Software Dependencies | No | The paper mentions software like LIBSVM [Chang and Lin, 2001], Language Tool, Stanford Core NLP [Manning et al., 2014], and SEMAFOR [Das et al., 2010] but does not provide specific version numbers for these tools. |
| Experiment Setup | Yes | We then train a linear support vector regressor (SVR) on these instances using the LIBSVM software package [Chang and Lin, 2001] with default parameters. |