24 Nov

#Armenia versus #Հայաստան versus #Армения

Twitter is a great platform, but for better or worse, it added support for non-Latin alphabets quite late. Today media.am had a great article on the use of the hashtags #Armenia, #Հայաստան (Armenia in Armenian script), and #Армения (Armenia in Russian).

I thought that I’d add an additional layer by doing a NodeXL hashtag analysis.

Here’s #Armenia


This is 2000 twitter users (a limit of 18,000 tweets) that used #Armenia in the last week. Group 1 are those that tweeted #Armenia but didn’t reply to or retweet anyone. Not surprisingly this is the majority of the use of this hashtag. They were mostly tweeting news stories about Armenia or were spammers.

The other groups though are quite interesting. They’re very tight clusters.
Group 2 seems to mostly be discussing the recent helicopter issue and features the popular Twitter user GoldenTent as well as the British Embassy in Armenia and the OSCE. Group 3 is talking about Armenia and Turkey and Kurds and seems to be mostly people in Turkey. Group 4 are talking about the city of Armenia in Colombia! Groups 6 and 10 are certainly Azerbaijanis based on the links that they share.

Here is #Հայաստան.


This contains only 90 Twitter users with tweets from the last week. There are almost no clusters as you can see. It does appear that these users are mostly writing in Armenian though.

And this is #Армения.


There were 162 Twitter users over the last week. Most were not replying or retweeting each others (see group 1) and their favorite links are usually instagram photos. Group 2, however, is considered a “broadcast” network where a lot of people were retweeting or replying to one user, SovietExplorer, a Russian news site. But Group 2 users didn’t have anything else in common other than a lot of tweets from Azerbaijani news sites. Group 6 is similar in a lot of posts from pro-regime Azerbaijani news sites.

I’d suggest that #Армения is a place where Azerbaijanis (also with Russian skills, obviously) are “seeking” to talk about Armenia.

27 Oct

Hashtag Shenanigans Again – Karabakh Edition, #azesaboteurs and #saveazehostages

It has been awhile since there has been hashtag shenanigans in the Caucasus. Some of the major hashtag shenanigans players fell out of favor. But this week things heated up again. I started seeing random odd tweets from accounts and upon clicking through it seemed like they were likely fake accounts – brand new, stock photos for the profile picture, few followers. These were usually in response to any criticism of Azerbaijan, but often with regard to Nagorno Karabakh.

This week the presidents of Armenia and Azerbaijan are meeting with French president François Hollande and last week the German Foreign Minister visited both countries. (John Kerry met with with last month, Putin in late summer, etc. etc.)

But perhaps of greater interest is that this week the authorities in Nagorno Karabakh put two Azerbaijani citizens on trial after they were caught crossing the border (a third was killed). The men say that they were going to visit relatives’ graves. The NK authorities say that the men killed a military officer. Azerbaijani authorities also note that NK has no right to hold a trial because it isn’t a recognized state.  RFERL story ArmeniaNow story Armenian RFERL

The other day I noticed a Twitter hashtag and Facebook group (with 8000+ “attending”) being promoted by the Azerbaijani ruling party’s youth wing. These sort of rallying around the flag issues are always interesting to me and I was a little surprised to see a lot of my oppositionally-minded Facebook and Twitter friends following this hashtag: #SaveAzeHostages. There was even a photo hashtag meme thing happening.


Meanwhile, Armenians created their own hashtag: #AzeSaboteurs.

This quickly turned into a hashtag battle.

It appears as if there are some fake accounts of some sort. For example, look at all these duplicate tweets (red/pink = duplicate).


Users zuma885, oqtay88, raminka10, and maxelmira, for example, have a lot of repeated tweets. And let us look at when these four joined Twitter! (Plus they use stock photos.)

These users showed up in both #SaveAzeHostages and #AzeSaboteurs.


Here’s the network analysis for #Azesaboteurs. It seems like Azerbaijanis may have “taken it over” but not to the extent that we saw in previous hashtags like #armvote13.


But look at #saveazehostages – what an interesting network analysis!


That center is the official Twitter account of the ruling party’s youth wing. Another power player is msabina34 who seems to be most interesting in One Direction. She tweeted on this hashtag over 100 times in one hour! There are also a number of other fake accounts on this hashtag.

It isn’t too hard to buy fake twitter accounts, but I wonder if this is a worthwhile investment? It is so obvious. I guess that the person in charge of this (likely someone at that youth wing) wants to show that s/he is a really dedicated member?

As a side note:

But this went beyond hashtag shenanigans and turned into a DDoS war. Here’s a report on what Azerbaijani websites Armenians took down. I don’t have any reports from Azerbaijanis.

03 Aug

Social Media and Information Wars

The past few days have seen a notable increase in cease-fire violations on the frontline of Nagorno Karabakh, with deaths on both sides (although more Azerbaijani deaths).

I’m currently in Azerbaijan, but I keep an eye on both Azerbaijani and Armenian social media spaces and these are a few thing that have happened on the Internet in the past few days:

1. A simultaneous “hacking war,” where teams from both sides try to take down websites, especially news sites, in the other country. This is far from new. Check out this article on this topic from the year 2000! Samvel Martirosyan does a good job keeping track of these sort of attacks from the Armenian perspective. There has been a major increase in DDoS and other attacks in the last few days, including popular Azerbaijani tabloid Haqqin.az being taken down on Sunday morning, although by noon Baku time, it was back up.

BuFzHCSIAAQc6G1(When the hacking teams take the site over, the usually post a graphic with the attribution of the hacking team that took it down.)

2. Information about what is going on is spread via social media and inevitably, information is not sourced well on either side. Granted, media is so politicized in this region that individual social media posters sometimes are the only source of information on a given topic. However, it creates situations where a piece of information can spread via social media and become “truth” quickly. This is especially the case when it comes to the number of soldier deaths – an individual says “There were 2 soldiers shot in THIS PLACE today.” without any source attribution.  In my estimation, there are more “quasi” journalistic sources in Azerbaijan than in Armenia and in both countries they are sometimes the source of new information, but that’s just my impression.

A case of this is photos of tanks going to the front line. One social media based Azerbaijani news source posted a photo of tanks on a train that was widely liked and shared. Then another set of tank photos was shared via social media and originated from a Russian news site, but social media users demonstrated that the photos were old. While it is possible that tanks are going to the front and that some of the photos are recent, some are obviously not.

Another interesting case that came out was ANS-TV (Azerbaijani state TV)’s website posted that there were 51 casualties on the Armenian side, and cited an Armenian journalist named Eduard Abramyan’s (that’s an incorrect spelling of that surname, by the way) Twitter account. But savvy social media users saw that the ANS screenshot of the Armenian’s Twitter account had the “delete” button visible, and that can only appear if one is the OWNER of the Twitter account. (Photo via Meydan TV.)


3. Yet, it is understandable that people are getting news from social media. There was a lot of anger in Azerbaijan on Friday that the TV stations were not covering this major news story. (Photo via North Caucasus Caucus.)


But at the same time, I wonder if the local TV stations CAN actually cover this story “live” a la CNN. I don’t know if they have the authorization to do so, or the capacity. Plus, this isn’t like an earthquake where there is a public service need to inform people “live” about what is going on. I suspect that people are angry and they want to direct their anger somewhere and the local TV stations are an open target.

4. And then there is the demonstration of concern about this issue via social media. The initial reaction amongst Azerbaijani social media users, across the political spectrum, was a great deal of profile and cover photo changes (or Instagram photo posts) to a black ribbon, sometimes with the Azerbaijani flag. (Much of it is of a religious nature as well, as noted by North Caucasus Caucus on Twitter, who is also collecting a lot of the social media posts.)


ribbon2 remember

And beyond individuals, a number of the Facebook pages for Baku shops and eating establishments also started posting memorial graphics.

2014-08-02 20.31.402014-08-02 20.27.402014-08-02 19.51.102014-08-02 08.39.572014-08-01 11.54.512014-08-01 12.49.06

I guess if you’re a business, you’re damned it you, damned if you don’t.

For both individuals and businesses, there is a great deal of social pressure to demonstrate this sort of concern for this issue. I’m not saying that the concern is not genuine, but that social media encourages this sort of viral social pressure mentality. (See other cases of this here and here.) And this may be especially true in Azerbaijan where the demonstration of patriotism/nationalism is especially salient. (See here and here).

5. Twitter, unlike Facebook, is a space where Armenians and Azerbaijanis can “discuss” this issue and there certainly seems to be a lot of chatter, compared to the normal quantity of Twitter activity in both countries.

These are all the mentions of the word “Karabakh” on Twitter in the last 9 hours and the links between those users (note that most people in Armenia and Azerbaijan were sleeping at the time, so a lot of the posters are based in North America and Europe). Most mentions are disconnected from any other users – a lot of posting of news stories, basically. So this introduces a lot of people into a Twitter analysis who might just be posting a big news story “There is fight in Nagorno Karabakh” and they aren’t usually involved in regional Twitter stuff, nor are they engaging in any sort of discussion. This makes social media analysis tough!

Yet I have seen (not measured, but seen) an increase in random Armenians and Azerbaijanis responding to Twitter posts. I suspect that some (bored? young?) people do searches on Twitter to find people to “troll.”

In conclusion, this is a scary time and the escalation of violence is upsetting. Karabakh is a “frozen” conflict that is not actually frozen at all. And the Internet allows for information to be shared and disputed – which is both a good thing (more information from more sources may be good; and as more citizens have the ability to report on what they see, we know more about what is going on) and a bad thing (the ability to create false information, lack of attribution). Social media as a platform is similarly good and bad. People can demonstrate their concern for what is going on and discuss events, but also the harassment through hacking and trolling brings a lot of negativity to an already negative situation.

I wonder how the regimes themselves feel about all this social media information spreading. Are the regimes using it to their advantage or is it a dangerous unknown variable in the equation of battle?

There is a lot of other stuff going on in both states right now – Azerbaijan is in the midst of a human rights crackdown and Armenia is on track to become closer to Russia (who arms both sides of this conflict, FWIW). Escalations on the frontline may very well be a tool to distract citizens from other issues and rally them around a concern about “the enemy” – also this provides a good social pressure tool – “why are you so worried about X when our boys are being killed!?!”. And based on this social media analysis, people are very distracted by this escalation right now.







10 Jan

Facebook Friends analysis… So many options!


A Facebook plugin called Find Your Best Friend On FB has blown up this week.

I assume that it looks at private messaging frequency and frequncy of likes and comments and then pops out a photo of you with that person.

I don’t really like these apps because they’re scanning your private messages and that makes me nervous. I did install it, run it, and then remove it though. It did seem to be accurate.

But there are a ton of other fun ways to visualize and analyze your Facebook network.

The first one is from Wolfram Alpha which has a plug in that does a lot of analysis and visualization, although no “best friends.”

Instead it has social insiders (shares the most friends) and social connectors (bridges between groups (like my twin sister is a bridge between my family group and my schoolmates group).

You can also see what friends like your stuff the most and comments on your stuff the most. That was really interesting to me.

NameGenWeb is my favorite app for this but it is down right now.

TouchGraph is pretty good too and is fast and mostly focuses on the networks that people have put themselves in (usually university networks). It ranks your friends (on number of friends in common). This wasn’t very accurate for me because it looks at those university networks more strongly than other factors. So, for example, a lot of people I grew up with attended Michigan State University in the late 1990s. I also have a number of friends who I did not grow up with who went there for their PhDs in the mid-late 2000s. That doesn’t mean much in terms of the connections between these groups.

FriendsGraph is an interesting one but the visualizations are pretty boring.


I like to use NodeXL and its Social Media Importer, but using it requires you to download some programs, so this is not a quick click. But the analysis is much more detailed (I added the labels to this picture).




03 Dec

#putinout hashtag analysis

civilnet (Via Civilnet.am)

There is a series of protests in Armenia right now, focused on Vladimir Putin’s visit, but overall critiquing Armenia’s possibly joining the Eurasian Union.

Unzipped does a nice summary here.

Here’s my Dec 3 hashtag analysis that is the following relationships. It is basically Armenians (far left), foreigners (middle), and Russians (right).


You can see that pretty much everyone in Armenia that is on Twitter follows each other. 🙂

This is the Dec 3 reply analysis of the hashtag – much clearer about small communication networks.



Here is Dec 4 reply analysis.


10 Oct

Shenanagins again and again

I’ve written a lot about “winning” hashtags and how this is a boastful tool for organizations. Here’s what I wrote yesterday on it.

I’ve also written about trickery using Twitter. Here’s a good one.

But now I have a new story…

As I was going through the Azerbaijani election hashtags, I noticed something funny.


This is all the users that used the hashtag #azvote13 in the last day. Look at how many people opened a Twitter account on a few days in February of 2013.

So I went to look at what was going on in those days in February. Not only were those accounts made on the same day, but they were made within minutes of each other. (If they’re highlighted the same color, it was the same day.)



So it is possible that there was some sort of training where a bunch of people created Twitter accounts at the same time, so I looked more closely at these accounts made in these days in February.

They generally follow the same few people. They generally don’t tweet a lot. Most haven’t tweeted a lot until this election period. I did a search on Facebook for some of them and no Facebook profile came up for them. It is possible that these young people spell their name differently on Twitter and Facebook – but in general I find that most young Azerbaijanis pick one way to spell their names on social media and stick with it.

So what were they tweeting about on the #azvote13 hashtag?


The same thing. The red/pink tweets are duplicates and the orange highlight is the same time. And they are at the exact same minute.

While it is theoretically possible for tweets to happen at the same minute, 100 tweets at the same minute by accounts all made in the same few days within minutes of each other? That’s a stretch.

What is interesting about this:

– this is way less detectible than buying fake Twitter accounts because the names are Azerbaijani, the profile pictures look like Azerbaijani people, even though there isn’t a lot of evidence that they are real.
– it is entirely possible that these are real people, but regardless, I speculate that one person has all the passwords and probably uses a particular app to send out the tweets at the same time.
– UPDATE 10PM BAKU TIME: I did a quick reverse image search on some of the profile photos of some of the profiles. The photos are ones that people can download and use all over the place. Here are some links and PDFs 1 2 3. And here are the original profiles.
– so, nice try – this was certainly more sophisticated than prior attempts, but still a FAIL.

As always, I’m happy to share these Excel files for people to look themselves.

09 Oct

We are young, heartache to heartache we stand; No promises, no demands #azvote13

Hashtags are an interesting 21st century phenomenon. Hashtags are keywords to organize information to describe a tweet and aid in searching (Small, 2011). Hashtags also take on a symbolic quality and perhaps are an example of metacommunication. When I post a picture of parents recording every minute of a kids’ holiday concert, complaining about how this is ruining the experience, I can add a hashtag #guilty to metacommunicate that I too participate in this. Symbolic use of hashtags also could be a way to show solidarity with some larger Internet community. When one sees a hashtag on a written sign at a rally, it shows some sort of connection to others that recognize the hashtag.

With that being said, analysis of a hashtag is much easier when there is a time-bound event like a rally or a protest, or in the case of this analysis, an election. When a hashtag is proposed for an event or topic, the intention is for a community of users to share information with each other. But hashtags also serve other purposes during an event: they can promote the event itself (come to the protest! #protest); they can give locationally situated information (the police are at the west gate #protest), and allow for live reporting that may send the message out to sympathetic others or media (Earl, McKee Hurwitz, Mejia Mesinas, Tolan, & Arlotti, 2013; Penney & Dadas, 2013).

For an election event in particular, a hashtag can serve as a means to share information, live report, and also to report possible fraud or violations. In the lead up to an election, a hashtag can be used for information dissemination.

With all that being said, hashtags also serve a boastful purpose. When a hashtag “trends” – it is noted by Twitter as being popular at a particular time. Users want a hashtag to trend to gain visibility and attention (Recuero & Araujo, 2012). While occasionally hashtags trend organically, it is much more common that hashtags are artificially pushed to the trending list (Recuero & Araujo, 2012).

This quantification of social media viability is very attractive. It allows a group to “prove” that it has a lot of support, even if it is artificial.

So, with that being said, let’s have a look at the hashtags for the 2013 Azerbaijani presidential election. (Here’s a pre-election report to give you a sense of the background).

A few different hashtags have emerged in the run up to the election. #azvote13 as well as #azvote2013, #secki2013, #sechki2013, and #besdir (the slogan of the main opposition candidate) are some of the most popular. I’ve been archiving #azvote13 and #secki2013 and #besdir for over a month.

It is important to note that pro-governmental forces have been engaging in some serious hashtag shenanigans all this year – hijacking hashtags thus rendering them useless for organizational purposes, amongst other things. However, in the past month or two, opposition youth have gotten more active on Twitter and seemed to take a stronger hold at different points recently. (Stronger hold = were the largest users of the hashtag.) There were a few alternative hashtags that emerged to make fun of a candidate, but overall the type of behavior that was seen earlier this year seems to have calmed down.

All this hashtag battling seems a little silly to me. But as noted before, hashtags are not just about being an information source, but about sending a signal. Thus, that tagging (to one’s own followers) is showing that “I’m talking about the election.”

Here’s some guidance on understanding these charts. One way to understand who a cluster is is to look at the links that they most frequently share and the source of those links. Are they tweeting from state news sources or opposition news sources, for example?

Here are the hashtags tracked over a long period of time:

#secki2013 – for the last month or so

The largest group of users of this hashtag were not connected to other people. But the second largest group was centered around opposition-leaning Hebib Muntezir (who has long been an information source for Azerbaijani news, lives outside of Azerbaijan, and currently works for an opposition-leaning sat/internet TV channel). Muntenzir’s following is quite interesting – the people around him aren’t heavily linked with each other. And they are receiving information from him, not vice versa – implying that he is an information source rather than a back-and-forth chatter.

Group 3 is the next largest group and it is the pro-government youth. They’re fairly interwoven with each other, and it centers on Rauf Mardiyev, the chairman of IRELI. You can see that his followers, like Muntenzir’s are a little distant from him. He is more a source of information than a chatter.

Group 4 is a very tight cluster – this is all the oppositionists. This is the first time that I’ve seen them all together in a cluster like this. Usually they subdivide into smaller clusters, but with links between them. There is a lot of communication between these people and a lot of efficient information sharing.

Group 5 contains the most active Twitter chatters of the opposition youth. They ended up in their own cluster probably because of how frequently they talk with each other.

#azvote13 – for the last month or so

The largest group here is the pro-government youth organization in Group 1. They’re pretty tight, although you can see some distant subclusters.

Group 2 is all the foreign media and NGOs that was covering the election as well as Azerbaijanis and foreigners that “affiliate” with these organizations.

Group 3 are the non-networked people.

Group 4 are the oppositionists and some media (usually when an opposition leader wrote a piece for a particular news organization and the tweet went out “@soandso’s article at @mediaorg”).

Group 5 is opposition youth – note the ties between group 4 and group 5.

Group 6 is muntenzir’s followers.

So what about election day itself?

#azvote13 for just election day

What’s interesting about election day itself isn’t so much the clusters as the content of what was tweeted – Lots of news coverage from every side. Also many instagram and Facebook photos of people voting, their ballots, and occasionally election violations.

So overall, Twitter is a battlefield. What is to be “won” is still not clear to me. I doubt this is a good use of anyone’s time to battle like this, but…

09 Oct

5pm Oct 9 hashtag update

More to come, but…

#azvote13 for last 48 hours

Group 1, the largest group, was unnetworked people.
Group 2 is a big cluster of foreign organizations and some individuals, Azerbaijani and foreign, associated with them. Seems like it was a lot of discussion and re-tweeting of Rebecca Vincent.
Group 3 is pro-government.
Group 4 is Azerbaijanis associated with the opposition.

#secki2013 for last 48 hours

#besdir for last 48 hours

#azvote2013 last 10,000 tweets