Outline
- Abstract
- Keywords
- 1. Introduction
- 2. Ego Network Structural Analysis in Online Social Networks
- 2.1. Ego-Network Model
- 2.2. Ego-Network Structure in Osns
- 2.2.1. Dunbar’s Number in Online Interactions
- 2.2.2. Hierarchical Structure of Ego Networks in Osns – Number of Circles
- 2.2.3. Hierarchical Structure of Ego Networks in Osns – Sizes and Frequencies of the Circles
- 2.2.4. Evolution of Online Ego-Network Structural Properties over Time
- 3. Information Diffusion Models in Osns
- 4. Combining Ego-Network Structural Properties and Information Diffusion Models
- 4.1. Analysis of Information Diffusion at the Ego Network Level – Single Hop Diffusion
- 4.2. Diffusion Models Based on Ego Network Structure – Multi-Hop Diffusion
- 4.3. Impact of Trusted Relationships and Ego Network Layers on Information Diffusion in Complete Osns
- 4.3.1. Network Coverage
- 4.3.2. Average Path Length
- 5. Conclusion and Future Research Directions
- Acknowledgments
- References
رئوس مطالب
- چکیده
- 1.مقدمه
- 2. تحلیل ساختاری شبکه فردی در شبکه های اجتماعی آنلاین
- 1.2 مدل شبکه فردی
- 2.2 ساختار شبکه فردی در شبکه های اجتماعی آنلاین
- 1.2.2 عدد دونبار در تعامل های آنلاین
- 2.2.2 ساختار سلسله مراتبی شبکه های فردی در شبکه های اجتماعی آنلاین-تعداد دایره ها
- 3.2.2 ساختار سلسله مراتبی شبکه های فردی در شبکه های اجتماعی آنلاین-اندازه ها و فراوانی دایره ها
- 4.2.2 تکامل ویژگی های ساختاری شبکه فردی در گذر زمان
- 3. مدل های پراکنش اطلاعات در شبکه های اجتماعی آنلاین
- 4. ترکیب ویژگی های ساختاری شبکه فردی و مدل های انتشار اطلاعات
- 1.4 تحلیل انتشار اطلاعات در سطح شبکه فردی- انتشار تک پرش
- 2.4 مدل های انتشار بر اساس ساختار شبکه فرد- انتشار چند پرشی
- 3.4 تاثیر روابط مورد اعتماد و لایه های شبکه فردی بر پراکنش اطلاعات در شبکه های اجتماعی آنلاین کامل
- 1.3.4 پوشش شبکه
- 2.3.4 طول مسیر متوسط
- 5. نتیجه گیری و جهت گیری های تحقیقات آتی
Abstract
Ego networks models describe the social relationships of an individual (ego) with its social peers (alters). The structural properties of ego networks are known to determine many aspects of the human social behavior, such as willingness to cooperate and share resources. Due to their importance, we have investigated if Online Social Networks fundamentally change the structures of human ego networks or not. In this paper we provide a comprehensive and concise compilation of the main results we have obtained through this analysis. Specifically, by analysing several datasets in Facebook and Twitter, we have found that OSN ego networks show the same qualitative and quantitative properties of human ego networks in general, and therefore that, somewhat counter-intuitively, OSNs are just “yet another” social communication means which does not change the fundamental properties of personal social networks. Moreover, in this paper we also survey the main results we have obtained studying the impact of ego network structures on information diffusion in OSNs. We show that, by considering the structural properties of ego networks, it is possible to accurately model information diffusion both over individual social links, as well at the entire network level, i.e., it is possible to accurately model information “cascades”. Moreover, we have analyzed how trusted information diffuses in OSNs, assuming that the tie strength between nodes (which, in turn, determines the structure of ego networks) is a good proxy to measure the reciprocal trust. Interestingly, we have shown that not using social links over a certain level of trust drastically limits information spread, up to only 3% of the nodes when only very strong ties are used. However, inserting even a single social relationship per ego, at a level of trust below the threshold, can drastically increase information diffusion. Finally, when information diffusion is driven by trust, the average length of shortest paths is more than twice the one obtained when all social links can be used for dissemination. Other analyses in the latter case have highlighted that also in OSNs users are separated by about 6 (or less) degrees of separation. Our results show that when we need trustworthy “paths” to communicate in OSNs, we are more than twice as far away from each other.
Keywords: Dunbar’s number - Ego networks - Information diffusion - Online Social Networks - Tie strength5. Conclusion and future research directions
In this paper we have presented our most recent work on the characterization of the structural properties of social relationships in OSNs, and how they depend on human cognitive and time constraints. From our analyses, we have seen that the properties of ego networks in OSNs are compatible with those found in offline environments. This indicates that the hierarchical structure of concentric layers of alters around the ego is consistent among different social environments, and is not influenced by the use of a particular communication medium. This is a clear indication that human cognitive and time constraints shape social relationships not only in offline environments, but also in OSNs, in contrast to the conventional wisdom that OSNs are able to improve our social capacity and allow us to maintain a much larger number of relationships than is possible “offline”.
In addition, we have shown that the structural properties of online ego networks can be used to understand in detail the process of information diffusion in OSNs, and to create more accurate predictive diffusion models. Tie strength is highly correlated with the amount of information that flows through each link, even though the correlation is higher for inner layers. This is consistent with the well-known Granovetter’s results, that showed that strong ties can carry a significant amount of information, although weak ties are also important for acquiring diversity of information (confirmed in our findings by the comparatively lower correlation between tie strength and amount of information). Moreover, properties of ego networks are highly correlated with the depth and size of information cascades originated from the ego. Finally, we have assessed the impact of tie strength on the diffusion of trusted information, showing that the well-known result about six (or fewer) degrees of separation does not hold when information can flow only over social relationships above a certain level of trust.