This paper develops a novel unsupervised algorithm for belief representation learning in polarized networks that (i) uncovers the latent dimensions of the underlying belief space and (ii) jointly embeds users and content items (that they interact with) into that space in a manner that facilitates a number of downstream tasks, such as stance detection, stance prediction, and ideology mapping. Inspired by total correlation in information theory, we propose a novel Information-Theoretic Variational Graph Auto-Encoder (InfoVGAE) that learns to project both users and content items (e.g., posts that represent user views) into an appropriate disentangled latent space. In order to better disentangle orthogonal latent variables in that space, we develop total correlation regularization, PI control module, and adopt rectified Gaussian Distribution for the latent space. The latent representation of users and content can then be used to quantify their ideological leaning and detect/predict their stances on issues. We evaluate the performance of the proposed InfoVGAE on three real-world datasets, of which two are collected from Twitter and one from U.S. Congress voting records. The evaluation results show that our model outperforms state-of-the-art unsupervised models and produce comparable result with supervised models. We also discuss stance prediction and user ranking within ideological groups.
翻译:本文开发了一个在两极化网络中进行信仰代表学习的新颖且不受监督的算法,该算法(一) 揭示了基础信仰空间的潜在维度,(二) 将用户和内容项目(与它们互动的)联合嵌入该空间,从而便利一系列下游任务,如定位检测、姿态预测和意识形态映射等。在信息理论总体相关性的启发下,我们提出了一个新的信息-理论变异图自动编码(InfoVGAE),该算法学习将用户和内容项目(例如代表用户观点的日志)投入一个相分离的适当的潜在空间。为了更好地将用户和内容项目(与其互动的)嵌入该空间,我们开发了完全的关联性或多层潜伏变量,从而便利了各种下游任务,例如定位检测、姿态预测、姿态预测和意识形态的分布。 用户和内容的潜在代表可以用来量化其在意识形态上的精细度并检测/判断其立场。我们评估了拟议InfoVGAE在三个真实世界数据集上的绩效,其中两个模型是从推特上收集的,并且从U-S-chroforstal的模型中展示了我们的排序中展示结果。