Dataset for "Neural embedding of beliefs reveals the role of relative dissonance in human decision-making".

This project contains the dataset used to generate the results of the study <b><i>"Neural embedding of beliefs reveals the role of relative dissonance in human decision-making"</i></b> (arXiv:2408.07237).<br>Authors: Byunghwee Lee¹, Rachith Aiyappa¹, Yong-Yeol Ahn¹, Haewoon Kwak¹, Jisun An¹¹ _{Center for Complex Networks and Systems Research, Luddy School of Informatics, Computing, and Engineering, Indiana University, Bloomington, Indiana, USA, 47408}<br><b>DDO_dataset.zip (</b>original Debate.org dataset<b>)</b><br>This archive contains the original raw Debate.org dataset, which was obtained from the publicly accessible website (https://esdurmus.github.io/ddo.html), maintained by Esin Durmus [1,2]. <b>All credit for this dataset belongs entirely to the original authors</b><b>, Esin Durmus and Claire Cardie.</b> We do not claim any authorship or modifications to this dataset. It is provided here solely for reproducibility and reference in our study.<br><br>The dataset includes the following three files:<br><br>- <b>debates.json</b>: This JSON file contains a Python dictionary that assigns a <i>debate name</i> --- a unique name for each debate --- to debate information<br>- <b>users.json</b>: This JSON file includes a Python dictionary containing user information<br>- <b>readme.md</b> file from the authors (Esin Durmus and Claire Cardie)<br><br>When using this dataset, please reference Debate.org and cite the following works:<br><br>[1] Esin Durmus and Claire Cardie. 2019. A Corpus for Modeling User and Language Effects in Argumentation on Online Debating. <i>In Proceedings of the 57th Conference of the Association for Computational Linguistics</i>. Florence, Italy. Association for Computational Linguistics.<br><br>[2] Esin Durmus and Claire Cardie. 2018. Exploring the Role of Prior Beliefs for Argument Persuasion. <i>In Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL).</i><br><b>df_ddo_including_only_truebeliefs_nodup(N192307).p</b><br>This file contains a pre-processed dataset used in our project (arXiv:2408.07237). The dataset includes records of user participation in debates (both as debaters and voters) as well as voting records across various debates. The belief triplet dataset used for fine-tuning a Sentence-BERT model was generated based on this pre-processed dataset. Detailed explanations of the pre-processing procedure are provided in the Methods section of the paper.<br><br>When using this pre-processed dataset, please cite the following reference (in addition to the two papers mentioned above):<br><br>[3] Lee, B., Aiyappa, R., Ahn, Y. Y., Kwak, H., &amp; An, J. (2024). <i>Neural embedding of beliefs reveals the role of relative dissonance in human decision-making.</i> arXiv preprint arXiv:2408.07237.<br><br><b>model_full_data.zip</b><br>This zip file contains five fine-tuned S-BERT models trained using a 5-fold belief triplet dataset. After unzipping the files, users can import the models using the 'sentence_transformers' Python library (https://sbert.net/).<br>

本项目包含用于生成研究《*信念的神经嵌入揭示相对认知失调在人类决策中的作用*》（arXiv:2408.07237）结果的数据集。作者：Byunghwee Lee¹、Rachith Aiyappa¹、Yong-Yeol Ahn¹、Haewoon Kwak¹、Jisun An¹¹ _{美国印第安纳大学布卢明顿分校信息学、计算与工程学院复杂网络与系统研究中心，印第安纳州47408}<br><b>DDO_dataset.zip（原始Debate.org数据集）</b><br>此存档包含原始Debate.org数据集，该数据集取自Esin Durmus维护的公开可访问网站（https://esdurmus.github.io/ddo.html）[1,2]。本数据集的全部著作权归原作者Esin Durmus与Claire Cardie所有，我们未主张对该数据集的任何著作权或修改权，此处仅为复现本研究及参考用途而提供。<br><br>本数据集包含以下三个文件：<br><br>- <b>debates.json</b>：该JSON文件采用Python字典格式，将每个辩论的唯一名称——<i>辩论名称</i>——映射至对应的辩论信息<br>- <b>users.json</b>：该JSON文件采用Python字典格式，包含用户相关信息<br>- <b>readme.md</b>：原作者Esin Durmus与Claire Cardie编写的说明文档<br><br>若使用本数据集，请引用Debate.org，并参考以下文献：<br><br>[1] Esin Durmus、Claire Cardie. 2019. 用于建模在线辩论中用户与语言影响的论证语料库. <i>见第57届国际计算语言学协会大会论文集</i>，意大利佛罗伦萨，国际计算语言学协会.<br><br>[2] Esin Durmus、Claire Cardie. 2018. 探索先验信念在论证说服中的作用. <i>见北美计算语言学协会会议：人类语言技术（NAACL）论文集</i>.<br><b>df_ddo_including_only_truebeliefs_nodup(N192307).p</b><br>该文件包含本项目（arXiv:2408.07237）中使用的预处理数据集。该数据集包含用户以辩手或投票者身份参与辩论的记录，以及跨辩论的投票记录。用于微调句子BERT（Sentence-BERT）模型的信念三元组数据集即基于此预处理数据集生成。预处理流程的详细说明见论文的方法部分。<br><br>使用此预处理数据集时，请额外引用以下文献（除上述两篇之外）：<br><br>[3] Lee, B., Aiyappa, R., Ahn, Y. Y., Kwak, H. & An, J. (2024). <i>信念的神经嵌入揭示相对认知失调在人类决策中的作用</i>. arXiv预印本arXiv:2408.07237.<br><br><b>model_full_data.zip</b><br>此压缩文件包含5个经过微调的句子BERT模型，均基于5折信念三元组数据集训练得到。解压文件后，用户可通过Python的`sentence_transformers`库（https://sbert.net/）导入这些模型。