Presentation rutgers
Download as PPTX, PDF1 like287 views
The document discusses a field experiment conducted using a Facebook-like social network platform called MyTito installed at a high school in Siena, Italy. Over three months, 253 of the school's 1,600 students actively used MyTito. The network showed partitioning between classes with no evidence of segregation. Preliminary analysis found the network was not fully connected and stopped at non-reciprocated friendships. Further analysis of message content confirmed when users focused on particular topics. The experiment showed promising early results but would benefit from connecting to other social networks and more in-depth semantic analysis of message texts.
Presentation rutgers
- 1. Guido Borà guido.bora @ unisi.it University of Siena 2013 Siena Workshop @ Rutgers
- 2. 04/05/2013 There is a considerable literature based on Social Networks sites (SNs) where the objective is to study the effects of interactions between users, for example friendship, classmate, roommate and tastes. (Lewis and Kaufman et al. 2008) (Jaxkson and Pin et al. 2009) According to the seminal literature (Lazarsfeld and Merton 1954, McPherson et al. 2001) individuals with similar attributes tend to end up together, this phenomenon is known as homophily which is Summarized by the famous proverb "Birds of a feather flock together". These studies had the advantage of extracting hidden structures and complex social ties however is still unclear, the cognitive impact of communication in the interaction between individuals (Chelmis and Prasanna 2011) and what is its influence on some aspects of students' everyday life: friendships, performance, sport and segregation of certain minorities. 2013 Siena Workshop at Rutgers
- 3. 04/05/2013 » Organization of lotteries with rewards (beauty contest and quizzes) and analysis of results » Study of, as permitted by the Privacy Act ,of tobacco (or alcohol) consumption » Analysis of influences on sport » Study of users communication through the analysis of the texts of the messages and posts » Study of the influence of friendship on the social network on school performance » Study of segregation (32% of the students are children of immigrants) 2013 Siena Workshop at Rutgers
- 4. 04/05/2013 » Attributes in the explanation of social behavior » Availability of longitudinal data » Possibility to collect dynamic data on multiple social relationships » Control effects of segregation » Text mining for semantic analysis and impact of content analysis 2013 Siena Workshop at Rutgers
- 5. In December 2012 we installed at Tito Sarrocchi , an high school of Siena, a platform Facebook like (based on the Elgg open source software) named MyTito. The school currently has 1,600 students. All students are registered by default. After 3 months, active users to the social network are 253. N 59 49 59 42 38 253 04/05/2013 Class I II III IV V Total Mean Age 2013 Siena Workshop at Rutgers % 23% 19% 23% 17% 15% 100% 15,4 Gender Immigrants 67 9 26,5% 29,0% Male 186 22 % Total 73,5% 253 71,0% 31 Female %
- 6. 04/05/2013 Sarrocchi representative students periodically organize meetings to explain to their colleague how MyTyto works. The peaks correspond to those days. Between February and March there was a slight growth in traffic. 2013 Siena Workshop at Rutgers
- 7. 04/05/2013 22% of users have only a friend on MyTito Social Network site SNs 2013 Siena Workshop at Rutgers
- 8. 04/05/2013 Density no loops = 0.02169245 2013 Siena Workshop at Rutgers
- 9. 04/05/2013 » It is not a complete network (in the original design we wanted to start with a complete network in which the friendship decayed if inactive after a period of time) » It is a multicentric network (groupings between classes) » The network stops with a friendship is not reciprocated » It is not a dense network 2013 Siena Workshop at Rutgers
- 10. 04/05/2013 » Communication barriers between the classes (despite the class representatives hoped that the social network could help new friendships) » No interaction with other social network sites » Facebook competition (Why do invent the wheel again?) » Declared uselessness of tool 2013 Siena Workshop at Rutgers
- 11. 04/05/2013 2013 Siena Workshop at Rutgers
- 12. 04/05/2013 We do not observe the phenomenon of segregation. » The Average Degree (AD) of the network is equal to 10.97 against the AD of nodes represented by the immigrants is 7.04 » Furthermore the AD of the partition of only immigrants is equal to 2.58 2013 Siena Workshop at Rutgers
- 13. 04/05/2013 2013 Siena Workshop at Rutgers
- 14. 04/05/2013 A 3-core analysis confirms the polarization into classes. Complete triads are mainly within the classroom 2013 Siena Workshop at Rutgers
- 15. 04/05/2013 2013 Siena Workshop at Rutgers
- 16. 04/05/2013 Analyze the influences of communication between users according to their level of awareness of communication. This communication lines are characterized by an increasing intensity of relationship which can be defined on three levels: » the first level, friendship, is characterized by a low cognitive involvement; » the level of private messages is characterized by an intermediate level of articulation; » the interaction on the wall is characterized by a high level of awareness especially if repeated over time. 2013 Siena Workshop at Rutgers
- 17. The objective of the PCA is to determine the relevant base of reference to represent the data and filter the noise hoping that this new base filters the noise and reveals previously invisible structures A. PCA is a linear transformation of the data: 1. 2. Minimize the redundancy measured by the covariance Maximize the information, as measured by the variance. B. The Principal Components (PC) are new variables that have the following properties: 04/05/2013 1. Each PC is a linear combination of the original variables 2. The PC are between them orthogonal, or are mutually uncorrelated, removing the redundant information, 2013 Siena Workshop at Rutgers
- 18. The PCA provides an alternative explanation of the observed variability with the advantage of describing the phenomena by unrelated dimensions and ordered in terms of their importance in explanation. This allows to: 1. 04/05/2013 2. To interpret the phenomenon through the new meaning assumed by the main components that have not been discarded to reduce the number of variables to consider, discarding the latest main components, which less contribute to observed variability 2013 Siena Workshop at Rutgers
- 19. 04/05/2013 Identification of three components in which to divide the text of messages where only one is clearly linked to a set of words related to the installation of the platform Once identified the components, they should be associated with a lexicon and an ontology 2013 Siena Workshop at Rutgers
- 20. 04/05/2013 This field experiment although preliminary, has promising results. Although in the absence of further qualitative information, data collected allow us to remarks: 1. The network is partitioned between classes (this is not sufficient to confirm Homopily) 2. Absence of segregation 3. Content analysis broadly confirmed when users are focused on particular topics 2013 Siena Workshop at Rutgers
- 21. 04/05/2013 » To develop APIs connection between MyTito SNs and Facebook or other SNs » To build a network with the texts of messages and posts. It requires a categorization with which to divide the network into semantic clusters » To overlay network users with the semantic one 2013 Siena Workshop at Rutgers