rumdetect2017

Datasets========The main directory contains the directories of two Twitter datasets: twitter15 and twitter16. In each directory, there are:- 'tree' sub-directory: This folder contains all the tree files, each of which corresponds to the tree structure given a source tweet whose file name is indicated by the source tweet ID. In the tree file, each line represents an edge given in the following format: ** parent node -> child node ** Each node is given as a tuple: ['uid', 'tweet ID', 'post time delay (in minutes)'] - label.txt file: This file provides the ground-truth labels of the trees in a format like: ** 'label:source tweet ID' - source_tweets.txt file: This file provides the source posts content of the trees in a format like: ** 'source tweet ID \t source tweet content' Note that constrained by the terms of Twitter service, we cannot contain the content of the rest of the tweets. Data users can obtain the sepcifics based on the provided tweet IDs and uids by their own.Feature description===================- Content features: uni-grams, bi-grams (presence/absence, binary)- User features: ** # of followers ** # of friends ** ratio of followers and friends ** # of history tweets ** registration time (year) ** whether a verify account or notReferences==========Substantial number of source tweets and their correspoding propagations trees were extracted based on two reference datasets described and released by the following works:- twitter15: @inproceedings{liu2015real, title={Real-time Rumor Debunking on Twitter}, author={Liu, Xiaomo and Nourbakhsh, Armineh and Li, Quanzhi and Fang, Rui and Shah, Sameena}, booktitle={Proceedings of the 24th ACM International on Conference on Information and Knowledge Management}, pages={1867--1870}, year={2015}}- twitter16:@inproceedings{ma2016detecting, title={Detecting Rumors from Microblogs with Recurrent Neural Networks}, author={Ma, Jing and Gao, Wei and Mitra, Prasenjit and Kwon, Sejeong and Jansen, Bernard J. and Wong, Kam-Fai and Meeyoung, Cha}, booktitle={The 25th International Joint Conference on Artificial Intelligence}, year={2016}, organization={AAAI}}If you use the datasets released hereby, please also cite:@inproceedings{ma2017detect, title={Detect Rumors in Microblog Posts Using Propagation Structure via Kernel Learning}, author={Ma, Jing and Gao, Wei and Wong, Kam-Fai}, booktitle={Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)}, volume={1}, pages={708--717}, year={2017}}数据集说明=================== 主目录包含两个Twitter数据集的文件夹：twitter15与twitter16。每个数据集目录下包含以下内容： - `tree`子目录：该文件夹存储所有传播树（propagation tree）文件，每个文件对应一条源推文（source tweet）的传播结构，文件名以该源推文的ID命名。传播树文件中，每行代表一条边，格式为：** 父节点 → 子节点 **。每个节点以元组形式表示：['用户ID（uid）', '推文ID', '发布时间延迟（单位：分钟）'] - label.txt文件：该文件提供各传播树的真实标签，格式为：** 标签:源推文ID ** - source_tweets.txt文件：该文件提供各传播树对应的源推文内容，格式为：** 源推文ID 源推文内容 ** 请注意，受Twitter服务条款限制，本数据集未包含其余推文的内容。数据使用者可自行通过提供的推文ID与用户ID获取相关细节。 特征说明=================== - 内容特征：一元语法（uni-grams）、二元语法（bi-grams）（以存在/不存在的二元特征形式表示） - 用户特征：** 粉丝数 ** 关注数 ** 粉丝与关注数之比 ** 历史推文数量 ** 注册年份 ** 是否为认证账号 参考文献=================== 本数据集的大量源推文及其对应传播树，均基于以下两项研究发布的参考数据集提取： - twitter15：@inproceedings{liu2015real, 标题={Twitter平台上的实时谣言辟谣}, 作者={Liu, Xiaomo、Nourbakhsh, Armineh、Li, Quanzhi、Fang, Rui、Shah, Sameena}, 会议名称={第24届ACM国际信息与知识管理大会论文集}, 页码={1867--1870}, 年份={2015}} - twitter16：@inproceedings{ma2016detecting, 标题={基于循环神经网络的微博谣言检测}, 作者={Ma, Jing、Gao, Wei、Mitra, Prasenjit、Kwon, Sejeong、Jansen, Bernard J.、Wong, Kam-Fai、Meeyoung, Cha}, 会议名称={第25届国际人工智能联合会议}, 年份={2016}, 主办机构={AAAI}} 若您使用本数据集，请同时引用以下文献： @inproceedings{ma2017detect, 标题={基于传播结构与核学习的微博帖子谣言检测}, 作者={Ma, Jing、Gao, Wei、Wong, Kam-Fai}, 会议名称={第55届计算语言学协会年会论文集（第1卷：长论文）}, 卷数={1}, 页码={708--717}, 年份={2017}}