Online social media has become an ideal place in spreading rumor events with its convenience in communication and information dissemination, which raises the difficulty in debunking rumor events automatically. To deal with such a challenge, traditional classification approaches relying on manually labeled features have to face a daunting number of human efforts. With the consideration of the realness of a rumor event, it will be verified and authenticated with multi-type individual information, especially with individuals’ emotional expressions to events and their own credibility. This paper presents a novel two-layer GRU model for rumor events detection based on multi-type individual information (MII) and a dynamic time-series (DTS) algorithm, named as MII–DTS-GRU. Specifically, MII refers to adopt the sentiment dictionary to identify fine-grained human emotional expressions to events and fuse with the individual credibility. Besides, the DTS algorithm retains the time distribution of social events. Experimental results on Sina Weibo datasets show that our model achieves a high accuracy of 96.3% and demonstrate that our proposed MII–DTS-GRU model outperforms the state-of-the-art models on rumor events detection.

中文翻译：

---

借助多种类型的个人信息，通过中文微博增强谣言事件检测能力

在线社交媒体以其便利的沟通和信息传播，已成为传播谣言事件的理想场所，这增加了自动揭穿谣言事件的难度。为了应对这样的挑战，依赖于手动标记特征的传统分类方法必须面对艰巨的人工工作。考虑到谣言事件的真实性，将使用多种类型的个人信息，特别是利用个人对事件的情感表达和自己的可信度，对谣言事件进行验证和认证。本文提出了一种基于多层个人信息（MII）和动态时间序列（DTS）算法的用于谣言事件检测的新型两层GRU模型，称为MII–DTS-GRU。特别，MII指采用情感词典来识别人类对事件的细粒度表达，并与个人信誉融合。此外，DTS算法保留了社交事件的时间分布。在新浪微博数据集上的实验结果表明，我们的模型达到了96.3％的高精度，并证明了我们提出的MII-DTS-GRU模型在谣言事件检测方面优于最新模型。