nealmgkr/tminer_hs

- This is a personal convenience copy of the binary Hate Speech (HS) dataset used in the T-Miner paper on defending against trojan attacks on text classifiers: https://arxiv.org/pdf/2103.04264.pdf - The dataset is sourced from the original paper\'s Github repository: https://github.com/reza321/T-Miner - Label mapping: - 0 = hate speech - 1 = normal speech - If you use this dataset please cite the T-Miner paper (see bibtex below), and the two original papers from which T-Miner constructed the dataset (see paper for references): ```@inproceedings{azizi21tminer, title={T-Miner: A Generative Approach to Defend Against Trojan Attacks on DNN-based Text Classification}, author={Azizi, Ahmadreza and Tahmid, Ibrahim and Waheed, Asim and Mangaokar, Neal amd Pu, Jiameng and Javed, Mobin and Reddy, Chandan K. and Viswanath, Bimal}, booktitle={Proc. of USENIX Security}, year={2021}} ```

原始信息汇总

数据集概述

• 数据集名称：Hate Speech (HS) 数据集
• 数据集来源：该数据集是从原始论文的Github仓库中获取的，原始论文为T-Miner。
• 数据集用途：用于防御文本分类器中的特洛伊攻击研究。

数据集标签映射

• 0：代表仇恨言论
• 1：代表正常言论

引用信息

• 引用论文：T-Miner: A Generative Approach to Defend Against Trojan Attacks on DNN-based Text Classification
• 作者：Azizi, Ahmadreza et al.
• 发表会议：Proc. of USENIX Security (2021)
• 引用格式：请参考提供的bibtex格式进行引用。