Digital War[Publisher’s Note: This article arrived yesterday August 5, 2014, but is being published on August 6, 2014]
Today at 8AM a cease fire has been declared in operation Protective Edge. The public discussion over the operation, however, will stay with us for a long time.
Public opinion is one of the key influence factors on countries and people behavior. In the digital and social networks world, that opinion might very rapidly turn itself into large-scale action (the ‘Arab Spring’, for example) , thus better understanding of it might shed some light on future trends and phenomena which are about to take place.
During armed conflicts, in addition to the traditional battle front, another front takes place in cyberspace, particularly if Israel is involved.
The MasterMineDS office is based in Tel Aviv. From here, a quick look at Facebook provides the common Israeli sense of solidarity with the military operation, proudly showing Israeli soldiers finding another rocket launcher or assault tunnel on one hand, and mutually helping each other at the home front between missile attack sirens, on the other hand.
However, while this is the situation in Israel, a brief view on TV channels, as well as news websites from all around the globe, actually shows very severe reactions, emotions and opinions against Israel.
Twitter is Left Behind
Israelis love social networks. They are highly active on Facebook, LinkedIn, and Instagram; WhatsApp and Viber were invented in Israel and are also very popular among its citizens.
But when it comes to Twitter, the amount of tweets per citizen is fairly low compared to the rest of the world, and it seems Twitter is being left behind in the Israeli arena.
With that in mind, we decided to research the global public opinion regarding operation Protective Edge as it reflects these days in Twitter, and by using sentiment analysis techniques find out whether Israel is regarded as a country who defends itself, or as an aggressor.
MasterMineDS’s Part – Tweet sentiment analysis
When you live in a country at war (like Israel), it is natural, although possibly incorrectly so, to feel as though the whole world is talking about the war going on in your backyard.
We at MasterMineDS are doing our part for the situation by trying to better understand the mood on Twitter regarding the on-going Israel-Gaza conflict from a quantitative, unbiased perspective.
We have decided to try answering the following questions through analyzing the data available from Twitter:
- Level of interest by country: What portion of the Twitter conversations are related to the Israel-Gaza operation in every country?
- Tweet sentiments: sentiment analysis will allow us to find the ratio of tweets supporting Israel’s activity to non-supporters in every country.
- Anomalies: Who are the anomalous users in different regions and what are they saying?
After setting up a server to collect millions of tweets per day, we started analyzing the data.
The analysis is based on roughly 17,500 daily Protective Edge related tweets, gathered for over a week on July 25-July 31, 2014, over 120,000 tweets in total.
Total tweeter conversations during that period stood on roughly 10,000,000 tweets per day.
Those numbers represent only a portion of the entire tweets database, due to capacity limits embedded in Twitter’s api. Nevertheless, the gathered sample consistently represents trends and proportion between countries.
First Goal: Level of Interest by Country
A point of interest in this research is the volume of the conversation around the world about the situation in Israel and Gaza.
We wanted to find out, per country, how many people are talking about the Israel-Gaza conflict through all the general tweets out there.
We have mapped the top 40 countries in aspect of the public conversation per country about the Israel-Gaza situation:
Who is interested in the Gaza situation?
We have organized the list by the amount of interest, Markers with higher then 1% of conversation are Big on the map, less are just a red dot.
As you can see, Pakistan leads the table, with 5.25% of the conversations in the Pakistani Twitter are about the Israel-Gaza conflict. Despite being a Muslim country, the numbers looked a bit high, therefore we decided to more deeply investigate it. See results below.
Next on the table are Israel’s neighbors Jordan and Lebanon, with 4.5% and 3.6% volumes, respectively. For those of you who were wondering, no, the conversations are not pro-Israeli.
Generally speaking, we noticed that the variance was high between the Arab countries that either contain a large population of Palestinian minorities or are known as radical vs moderate Arab countries.
One of the big surprises for us was Turkey, where the interest in the situation was around 0.45% despite the high volume of general tweets [Turkey contributes about 10% of the world’s tweets regarding Protective Edge]. As it seems, the people of Turkey love to tweet in general, yet not discussing the situation at the same extent. Are these the first signs of a difference in priorities between the Turkish government and its citizens? Only time will tell.
For the people of Israel who were planning on taking a trip to see the great Safari in Kenya, prepare yourself for plenty of questions from the locals. Our data shows as much as 3.33% of Kenya’s Twitter conversations are related to our subject.
South America, in general, seems to be worrying about other issues and despite some media noise from some countries over there, the current Middle East situation is not on their daily agenda.
The bottom line for the rest of the Western World is steady: less than 1% of the total conversations are related to the Israel-Gaza conflict.
Second Goal – Finding the Sentiment:
After measuring the extent of the conversation, we wanted to test the sentiment level of the tweets for\against Israel.
For that purpose, we have created an algorithm for finding the sentiment level, based on the severity of the tweet content.
Then, we created user segmentation, based on their amount of followers, tweets, sentiments and sentiment levels.
Finally, each user was assigned to one of 4 types: with\against Israel, and high\low social impact using Protective Edge related tweets.
In order to better understand the general feeling or pulse regarding the situation, we have created the following map, and pinned around 2,500 markers on it
– The map contains random tweets related to the conflict, and has a negative\positive sentiment.
– Pro Israeli sentiment is represented by green markers.
– Pro Palestinian sentiment is represented by the red markers
– The marker size represents the user’s involvement level
-The sentiment accuracy level for this map is 90%.
As you can see on the map, the USA is the only country that has a relatively high representation of the Israeli perspective by the users [around 30% of the tweets are in Pro-Israeli]. Other countries in the world may contain some Israel supporters, but they are quite a small minority.
Final Goal: Anomalies and Interesting Cases
The Twitter audience is generally not fond of users who are extreme in their views. Users who are tweeting radical content are bound to have fewer followers. With this in mind, we have decided to look for and try to track the users that express their tweets demonstrating extremist content, or exhibit properties such as frequency or having an interesting social root.
Here are some examples of our findings:
Zahid from Pakistan is flooding the stream with more than a 100 anti-Israeli tweets per day. Zahid managed to add 50 more followers to his list this week, with a total of 1,500.
The persistence of Zahid’s work in tweeting substantialy contributes to the extreme involvement levels in Pakinstan, a country that uses Twitter quite poorly.
A deeper analysis of the Growth Hacking Techniques used by Zahid can be found will be published soon.
Few Vs. Many
A few brave users can be found at the heart of some Arab countries, tweeting in favor of the Israeli side. For their own safety, and despite the great amount of appreciation we have for those users, we have decided not to reveal their user names. What we can say, however, is that it seems that some of those users are originally from Europe and are currently in those countries for work related reasons.
The first user is located in Kuwait, and he is calling on the Israeli government to stay strong.
Another user is located in Turkey. His tweets are arguing that the Hamas is a guerrilla organization who threatens journalists not to expose any violent actions it turns against the people of Gaza.
For the Attention of the Israeli ambassador in the UK:
The user Tony Huges is busy these days promoting an e-petition calling to expel Israeli ambassador Daniel Taub from the UK to his 1,500 followers.
Our findings show, that while the western countries are not extremely interested in the conflict, the amount of users who condemn Protective Edge operation is dramatically higher than those who support the Israeli side, legitimizing pretty severe messages through the web against the state of Israel.
As it has been demonstrated in this article, a deep sentiment analysis of social network data, such as Twitter, could lead to very interesting insights of global public opinion.
Intelligent use of some the of findings – in this case by Israeli foreign affairs officials and others – could help in engaging more people to help balancing the world’s public opinion, both during the fighting and after the cease fire.
About The writers:
MasterMine Data scientists is located in Tel Aviv, Israel and specializes in data science and analytic solutions