Skip to main content
Search form
  • 00:05

    [Anatoliy Gruzd Discusses Social Media Data& Social Network Analysis]

  • 00:09

    ANATOLIY GRUZD: Hello, I'm Anatoliy Gruzd.I'm a social media researcher.[Anatoliy Gruzd, PhD, Canada Research Chairin Social Media Data Stewardship,Ryerson University]I study how social media changes the way peopleand organizations communicate, collaborate,and share information.Well, it has a very huge impact on all aspects of our life.And so to understand how it really impacts our lives,

  • 00:29

    ANATOLIY GRUZD [continued]: I'm really interested in looking at studying the digital tracespeople leave on social media websites.And so one of the particularly useful techniquesthat we use in the Social Media Labwould be analyzing data from social media usingsocial network analysis.When we say we're studying social media data usingsocial network analysis, or SNA, it

  • 00:51

    ANATOLIY GRUZD [continued]: means we're trying to represent a data setin the form of a graph where nodes will be social mediausers, usually.And connection can represent different typeof relationships.And in fact, social media data is inherently social,because it's being generated by usersand it includes information about who'sconnected to whom, who talks to whom, who likes whose messages.

  • 01:14

    ANATOLIY GRUZD [continued]: And so knowing this, we can then identify if thereare subgroups in a group.We can identify who is more or less influential, and why.Well, really, in fact, any kind of datacan work with social network analysis.The fact that as researchers we sometimesdon't have access to all kinds of social and income data,and so most of us in the social media research area

  • 01:36

    ANATOLIY GRUZD [continued]: would be working with communication networks.So those are usually available for usto automatically collect.And specifically, when we're talking about communicationnetworks, the data can come from public forums, Facebook grouppages.It can come from comments sections on YouTube videos.

  • 01:57

    ANATOLIY GRUZD [continued]: It can come from blogs linking to each other or commentsthat you leave on Twitter.So it can come from any platforms.But the idea is that essentially, youneed to have information about who is replying to whomor who is liking whose content, so that youcan use those interactions to connect individuals.

  • 02:18

    ANATOLIY GRUZD [continued]: And usually, if you observe the data long enough,then you can start seeing patterns.The more people engage with each other,you can then extrapolate to say theyhave stronger relationships and connections over time.So social network analysis is really greatto answer a community of related questions.So let's say you have a group of users joining

  • 02:38

    ANATOLIY GRUZD [continued]: a website like Reddit, and they arediscussing topics like academia, or politics.And you want to know whether thereis a really strong sense of community among group members.So by representing their conversationsin the form of a graph, in the form of a network,and applying social network analysis,you actually can measure group cohesion.You can look for group leaders.

  • 02:59

    ANATOLIY GRUZD [continued]: And what's especially powerful, if youlook at the changes in the network over time,because you can start predicting whether a group willbe successful in the future or whether it'sgoing to just disappear.So there are quite a few challenges.The data access would be one of them.And even though the data we're talking about in our research

  • 03:20

    ANATOLIY GRUZD [continued]: would be publicly available, sometimes it'schallenging to collect it because wehave to constantly negotiate different rulesand techniques by working with platforms like Facebook,Twitter.And their rules and techniques, whatthey give us in terms of data points, changes all the time.So even though once the network is ready,It's quite straightforward to analyze it

  • 03:40

    ANATOLIY GRUZD [continued]: because there are established techniques and softwareproducts out there to analyze networks,just to get to that point it may take us some time to figure outA, what groups you're going to study,what access you have to those groups, canyou collect the data automatically,how do you represent the interactions as a network?So those would be related to data collection, some

  • 04:03

    ANATOLIY GRUZD [continued]: of the challenges.So once you've got the network, then youcan look at the group level measures.For example, density centralization,modularity, clustering, will all tell you a little bit moreabout the group overall, a network overall.And then you can also zoom in and studyparticular individuals in the group.

  • 04:25

    ANATOLIY GRUZD [continued]: So those would be node level or individual level metrics,like centrality, for example.A lot of people are studying influenceand how do we measure influence online and offline.They would be using centrality measures,like degreed centrality, betweenest centrality,closeness centrality.So those are indicators of potential influence

  • 04:47

    ANATOLIY GRUZD [continued]: that somebody might have in a group.But especially powerful if you are able to also connectthese data-driven observations and SNA-drivenmetrics with interviews, with some kind of qualitative data.Some of our work will involve actually locatingpotential influencers in a group and interviewing them

  • 05:07

    ANATOLIY GRUZD [continued]: about their involvement in a particular group.So this way, you can apply a mixed method approachto studying communities online.So one of the challenges of applying social networkanalysis to online communities isthat your results may be differentdepending on when you collected your data,because a lot of people join a group for some time

  • 05:28

    ANATOLIY GRUZD [continued]: and then they leave, or maybe they join but they never post.So if you only collect data during a one month period,you may be missing a lot of other interactions.So one of the suggestions that I would proposeis make sure that you have latitudinal data, whereyou have there is a snapshots of a group,and that you can then compare over time, changes over time.

  • 05:50

    ANATOLIY GRUZD [continued]: And that will actually allow you to seeare there any participants who join the group,but they did not post to the group right away,but maybe they became more comfortable with the grouprules and norms and contributed later on in the process?So really including this time dimension in network analysiswould be really crucial in building a solid foundation

  • 06:12

    ANATOLIY GRUZD [continued]: for your results.

Video Info

Publisher: SAGE Publications Ltd

Publication Year: 2019

Video Type:Interview

Methods: Social network research, Social media research, Computational social science

Keywords: blogs; communication networks; Facebook; group dynamics; internet data collection; Reddit; Social media; Social network analysis; Social network analysis and issues; Social networking communities; Twitter; YouTube ... Show More

Segment Info

Segment Num.: 1

Persons Discussed:

Events Discussed:



Anatoliy Gruzd, PhD, Canada research chair in Social Media Data Stewardship at Ryerson University, discusses the use of social network analysis (SNA) to study online social media communities, including types of data, and challenges like data access and shifting group dynamics.

Looks like you do not have access to this content.

Anatoliy Gruzd Discusses Social Media Data & Social Network Analysis

Anatoliy Gruzd, PhD, Canada research chair in Social Media Data Stewardship at Ryerson University, discusses the use of social network analysis (SNA) to study online social media communities, including types of data, and challenges like data access and shifting group dynamics.

Copy and paste the following HTML into your website