论文标题
签名的自我网络模型及其在Twitter上的应用
Signed ego network model and its application to Twitter
论文作者
论文摘要
自我网络模型（ENM）描述了个人如何以同心圆（通常为五个）减少亲密关系来组织其社会关系，并且在离线和在线的社交网络中几乎发现了它几乎无处不在。 ENM在相互作用频率方面衡量了同行之间的平局强度，这很容易衡量，并为培养这种关系所花费的时间提供了良好的代理。但是，签名网络分析的进展表明，正相关和负面关系在网络动态中起着截然不同的作用。因此，这项工作旨在调查包括签名关系时的ENM。本文的主要贡献是双重的：首先，一种使用情感分析签署个人之间关系的新方法，其次，研究了签名的自我网络（具有签名连接的自我网络）的属性。然后为八个不同的Twitter数据集的用户提取签名的自我网络，该数据集由专业用户（例如记者）和通用用户组成。我们发现，在所有类型的用户的自我网络的活跃部分中，负面链接代表过多，这表明Twitter用户倾向于定期与负面连接互动。此外，我们观察到，在专业用户的自我网络圈子中，负面关系主要是主要的，这暗示了这类用户的极度两极化的在线互动。此外，在ENM的较亲密级别的记者中发现了负面关系，而他们的百分比在其他Twitter用户的圈子中稳定
The Ego Network Model (ENM) describes how individuals organise their social relations in concentric circles (typically five) of decreasing intimacy, and it has been found almost ubiquitously in social networks, both offline and online. The ENM gauges the tie strength between peers in terms of interaction frequency, which is easy to measure and provides a good proxy for the time spent nurturing the relationship. However, advances in signed network analysis have shown that positive and negative relations play very different roles in network dynamics. For this reason, this work sets out to investigate the ENM when including signed relations. The main contributions of this paper are twofold: firstly, a novel method of signing relationships between individuals using sentiment analysis and, secondly, an investigation of the properties of Signed Ego Networks (Ego Networks with signed connections). Signed Ego Networks are then extracted for the users of eight different Twitter datasets composed of both specialised users (e.g. journalists) and generic users. We find that negative links are over-represented in the active part of the Ego Networks of all types of users, suggesting that Twitter users tend to engage regularly with negative connections. Further, we observe that negative relationships are overwhelmingly predominant in the Ego Network circles of specialised users, hinting at very polarised online interactions for this category of users. In addition, negative relationships are found disproportionately more at the more intimate levels of the ENM for journalists, while their percentages are stable across the circles of the other Twitter users
