• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site

Interdisciplinary internet studies

Priority areas of development: sociology
2015
The project has been carried out as part of the HSE Program of Fundamental Studies.

Project “Interdisciplinary internet studies” consists of three sub-projects: “Representations of social groups online and inequality”, “Online social networks and success of nascent entrepreneurs”, and “Development of new online packet processing algorithms”. Each sub-project has its subject, goals and empirical base.

Subproject “Representations of social groups online and inequality”

Research goal. This subproject aims at finding out how the structure of friendship networks in social network communities, and especially inequality parameters in these networks, are related to the goals and declared topics of these communities. In other words, the study seeks to find out whether different kinds of online communities generate different structures of inequality.

Empirical base of this subproject consists of public groups of three kinds in the Vkontakte social networking site. These groups were selected so as to not be online replicas of offline communities and organizations, they number to at least 4000 participants. The group types of interest include professional communities, fan communities, and social movements. We have studied 55 groups: 19 groups of the first and second type each, and 17 groups of the third type; in total, they numbered 827093 participants, 2091268 friend connections, and 318020 posts.

Results. We have shown that the purpose and topic of an online community significantly determine the social structure of this community, namely, they influence how a network of connections is formed between participants of the community. For instance, fan networks have lower density and larger numbers of i network components, which indicates a more fragmented network of friendship. We can conclude that participants of fan communities are least likely to use them to accumulate group social capital.

In professional communities, the dominating form of participation is inactive membership. We assume that this is due to the use of Vkontakte groups as content providers and means of declaring the users' professional affiliation rather than professional communication. On the contrary, members of social movement communities are mostly very active participants of group communication; the friendship networks are tighter than in other kinds of communities, and clear leaders measuered by the number of friends are observed (as evidenced by the Gini index with respect to centrality). This suggests that the direction towards collective action presumes active participation and the presence of efficient leaders, which causes such kind of structure for social movement communities.

Subproject “Online social networks and success of nascent entrepreneurs”

Research goal. The main goal of this subproject is to establish relation between the involvement of nascent entrepreneurs (startupers) in social networks and their other online characteristics, on the one hand, and their startups' success, on the other.

The empirical base for the second subproject consists of the following parts: (a) user accounts from the largest Russian online platform for startupers, Startuppoint, from the moment of its launch (2008) to the moment of data collection (September 2014), 3764 profiles in total with detailed information regarding their projects; (b) user accounts from the Russian Startup rating platform (RSUR), the only Russian online platform with public evaluation of the startups' investment attractiveness, for the second half of 2014 (i.e., the maximal possible amount available at the moment of data collection); this amounted to 1508 entrepreneurs with 1672 projects and their evaluations on a 1 to 10 scale along six different characteristics; (c) profiles of entrepreneurs from both platforms in the Vkontakte social network, 2587 profiles in total with all their data (metadata, posts on the walls, number of friends and so on).

Results. We obtained a portrait of an average startuper who is also a Vkontakte user; it is a 25-35 year old male with technical education who is doing development in the IT industry, lives in a large Russian city, and has a statistically significantly large number of friends, subscribers, and groups than average VKontakte user.

After analyzing the Russian Start-up Rating (RSUR), we show that the most important online characteristic of a startuper that influences a higher rating obtained by this startuper is the larger number of friends among other startupers; the total number of Vkontakte friends does not have a significant influence. Other factors are also present, but are not as influential. For instance, we have found that the more likes a page owner has for his or her comments relative to all likes to all comments on the page, the higher is the probability that he or she get a higher PR and financial rating. We have also established that predictors interact with each other in nontrivial ways. In particular, we have found that the number of friends among startupers is more important in Moscow than other parts of Russia; in different industry domains different characteristics become important, and the effects of online characteristics differ in different cities.

Subproject “Development of new online packet processing algorithms”

Research goal. The goal of this subproject is to develop new online algorithms for network packet processing, namely for network buffer management in various settings and study the quality of these algorithms. Network packets here are data blocks transmitted in computer networks, such as the Internet, in the form of electrical signals; in network switching, packets are stored, processed, and transmitted in buffers, which leads to important buffer management algorithms.

The empirical base for the third subproject consisted of synthetic network traces imitating the operation of a network switch with various buffer management policies for the buffers storing network packets; in these simulations, we simulated the properties of Internet traffic and tested the quality of different buffer management policies on this synthetic data. The traces were generated with 100 independent sources, each simulating an on-off Markov modulated Poisson process; the simulations lasted for 500000 rounds, and in every experiment we varied different parameters (intensity, maximal required processing, buffer size, number of cores etc.)

Results.   First, we have developed a new family of buffer management policies, namely lazy algorithms, and a new approach to evaluating their quality. We have proven lower bounds on the competitive ratio of PO and LPO online algorithms in a number of settings and upper bounds for the LPO algorithm; we have conducted a detailed experimental evaluation on synthetic traces that has supported our theoretic results. Second, in the field of network classifier optimization, we have introduced the novel notion of order-independent classifiers (with respect to the rules they consist of) and have shown that new fields in order-independent classifiers neither increase TCAM space nor decrease the performance of classifiers.

Recommendations for practical use. Buffer management policies developed and studied in the third subproject can be used in real life network processors (switches). Algorithms for classifier optimization developed in the third subproject can be used immediately before storing the resulting classifiers in TCAM memory. This work is becoming more and more relevant with the increasing popularity of the IPv6 protocol; our results can significantly simplify and reduce IPv6 classifiers.

Publications:


Мейлахс П. А., Рыков Ю. Г. Онлайновое сообщество СПИД-диссидентов в социальной сети «ВКонтакте»: структура и риторические стратегии // В кн.: XV апрельская международная научная конференция по проблемам развития экономики и общества: в 4-х книгах / Отв. ред.: Е. Г. Ясин. Кн. 3. М. : Издательский дом НИУ ВШЭ, 2015. С. 137-146.
Zheluk A., Quinn C., Meylakhs P. Internet search and krokodil in the Russian Federation: an infoveillance study // Journal of Medical Internet Research. 2014. Vol. 16. No. 9. P. e212. doi
Рыков Ю. Г. Сетевое неравенство и структура онлайн-сообществ // Журнал социологии и социальной антропологии. 2015. Т. XVIII. № 4. С. 144-156.
Semenov A., Natekin A., Nikolenko S. I., Upravitelev P., Trofimov M., Kharchenko M. Discerning Depression Propensity Among Participants of Suicide and Depression-Related Groups of Vk.com, in: Analysis of Images, Social Networks and Texts. 4th International Conference, AIST 2015, Yekaterinburg, Russia, April 9–11, 2015, Revised Selected Papers / Ed. by M. Y. Khachay, N. Konstantinova, A. Panchenko, D. I. Ignatov, V. Labunets. Vol. 542: Series: Communications in Computer and Information Science. Switzerland : Springer, 2015. Ch. 3. P. 24-35. doi
Semenov A., Natekin A., Nikolenko S. I., Upravitelev P., Trofimov M., Kharchenko M. Communications in Computer and Information Science, Vol. 542, Springer, 2015 Vol. 542: Analysis of Images, Social Networks and Texts. Fourth International Conference, AIST 2015. Springer, 2015. doi
Kogan K., Nikolenko S. I., Rottenstreich O., Culhane W., Eugster P. Exploiting Order Independence for Scalable and Expressive Packet Classification // IEEE/ACM Transactions on Networking. 2016. Vol. 24. No. 2. P. 1251-1264. doi