Social Media Data Analysis
Transkript
Social Media Data Analysis
Social Media Data Analysis Hüseyin Oktay University of Massachusetts, Amherst hoktay@cs.umass.edu Recep Erdem Erkul, Editor & Founder www.digital-government.net eerkul@digital-government.net (This article published on Informatics Association of Turkey Journal ) (http://www.bilisimdergisi.org/s153/ index.html?page=24 ) Second Generation Internet Web 2.0 (i.e,. second generation Internet) was coined in 2004 by O’Reilly Media[1]. Second-generation internet includes social networking sites, social media platforms and wikis—i.e. systems utilized by internet users for collaboration and sharing. Web 2.0 was about blogs and wikis in its early forms; however, moving forward, other web technologies emerged to improve Web experience. New internet technologies developed within such purpose enable visitors to participate with web sites and platforms, to collaborate with other users as well as easily share their ideas and thoughts [2]. Facebook, YouTube, MySpace, LinkedIn, Twitter, Wikipedia and Blog pages are examples of the most widely used Web 2.0 applications. These applications did not exist in our lives until 2004; however, now it is almost impossible to imagine our lives without them. To reiterate, content is created entirely by users in Web 2.0 and social media applications and individuals can interact with each other through these applications. This main feature of these applications transforms the power in content creation to an equal and participative model instead of an ‘elitist’ model of monopoly [3]. Perhaps as a result of this participative model, Web 2.0 and social media are being very actively used today. According to recently published studies, users around the world spend 1 out of total of 5 online-minutes in using the social media platforms [4]. In August 2011, 23 million different users spent a total of 45.3 billion minutes on the internet. With 13.1 billion minutes, users spent most of their time on Facebook [5]. More than 200 million tweets per day is shared on Twitter throughout the world [6]. More than 5 million of that tweets are sent in Turkish. With such evidence, society’s views, requests, and responses (or at least part of the society represented in social media), to the current events are reflected in social media. Comments on all kinds of topics and events; from a political event to TV series, or from the results of football matches to economic news and discussions are present in social media or blogs [7]. Moreover, state institutions—via e-participation efforts—or companies—via social marketing strategies—try to encourage citizens or consumers to share their views through social media [2]. In other words, social media becomes a platform which comments and ideas of thousands, perhaps millions of users on various events are taking place. There are thousands, maybe millions of different posts required to be read or summarized. Crimson Hexagon Advancements in artificial intelligence and computer technology enable us to index and summarize such extensive content generated on social media. For example, a social media analytics company, Crimson Hexagon [8] co-founded by a Harvard professor in Boston, created Foresight platform. Such platform can collect and automatically analyze posts—on a specific topic—created by users on Twitter, Facebook, YouTube and Blogs. An analysis report is created automatically within the platform. These reports, include proportions for custom generated categorized related to a specific topic. In addition, aggregate level demographic information and the popularity of users posting about these subjects (based on their Klout score) are automatically created. Arguably, users on social media may not be representative of the whole society. However, understanding opinion of part of the society that is represented on social media could be very useful to government agencies as well as different companies. Our examples will span a wide spectrum, from current global events to companies’ understanding consumer motivations, to people’s reaction on television programs and political events took place within the country. Case Studies Egypt and Libya The first example is about the analysis of the reaction of people in Egypt, Libya and Syria on twitter to Arab Spring. Different studies argued that people used social media as a coordination and communication platform during Arab Spring. In this article, we mention a study that analyses twitter content created in these countries before and after Arab Spring [14]. In Egypt, twitter users mainly generate posts expressing both their opinions about the current political situation and how the new government should be created. From February 2011 to June 2012, approximately 12 million tweets in Arabic and more than 1 million tweets in English are analyzed for this study. During this time-interval, conversations on Twitter, focused on forward-looking approach, which is about establishing new governmental institutions and elections (i.e., nation building) instead of backward-looking that is about revolution and the actual Arab turmoil. While 54% of tweets have focused on revolution between February 2011 and November 2011 elections, this percentage has decreased to 26% during the period after the elections until January 2012. In June 2012, this proportion has dropped to 10%. This topic was replaced by the conversation about establishing a new state. The conversation on Twitter about Libya’s was a little bit different than it is in Egypt. In September 2011, after the death of Kaddafi, twitter conversations are different depending on the language it is written (i.e., English or Arabic). Tweets in English talked about armed forces and armed conflict, on the other hand, tweets in Arabic were mostly about establishing a new state. Also Libya, did not seem to be moving on after Kaddafi—one third of the conversation in Twitter is still about the crimes during Kaddafi and his family’s ruling and the punishments they are exposed. Pope Election on Twitter and Facebook The second example includes the analysis of posts in Facebook and Twitter related to the election of Pope on 13th of March 2013. The reaction to the election is compared between the USA, UK and Argentin. In USA, while 21% of the conversation were positive, 8% of them were negative. In UK, on the other hand, the proportion of the positive and negative conversations were equal with 12%. Maybe not surprisingly, the proportions were higher than these two countries in Argentina: 30% positive and 18%negative. [16] Conversations done in USA were mostly (56%) just sharing the news without including any subjective opinion. 15% of the positive conversation is about the confidence and the excitement for the new Pope. 6%, on the other hand, is about the excitement for the Catholic Church in general. In UK, 51% of sharing content conversation included some humor sand sarcasm. 13% of the positive opinion in Argentina, the country of the new Pope included support and excitement, 10% expressed the national proud after an Argentinian to be elected for this duty. In this example, beside the ease of measuring the reactions for a current event by looking at social media, we can see that the used technology is language agnostic in its analysis. Valentine’s Day—States in USA Third example is about Valentine’s Day tweets in different states in the USA. By analyzing more than 2 million tweets shared in the first two weeks of February, the overall picture was as follows: 23% of them were neutral, 40% of them negative and 37% of them were positive. While the tweets with negative opinion were more than that of positive opinion in the first days of February, the daily percentage of the positive conversation became more than the negative conversation as the Valentine’s Day approached. [15] In U.S. the state where the most romantic conversation appeared via twitter was California famous for its perfect sunny weather through out the whole year. Wyoming and Texas were the runner-ups after California, and the remaining states for top 5 are Utah and Hawaii which is a popular honeymoon destination. The most negative conversation about Valentine’s Day were in South Dakota where usually severe weather conditions happen, the other states in top 5 for negative conversation were West Virginia, North Dakota, Rhode Island and Kentucky. The largest amount of tweets about Valentine’s Day are posted in New York, the state having the largest young and single population. Because of the tweets frequently showing an intention for finding a date, it seems the most convenient place for people to go not to be alone on Valentine’s Day was obviously New York. U.S Presidential Elections Another example is the third U.S. Presidential Debate on 22th October [9]. For this debate, in two days, approximately one million tweets and photo were shared in Twitter. Analysis of such posts revealed that, 37% of the users think Obama wins the controversy, on the contrary 35% of users think Romney wins. The remaining %27 think the result think it was draw. [10] Posts favoring Obama state that he is a better option for economy and foreign politics. Tax Increase in US The fourth example is the citizens’ opinions in social media about the US tax increase. 24% of the overall conversations in social media reflect the emotions of people (e.g., posts from people who are angry, anxious and disappointed). 36% of them mainly share news items about the issue. 22% share ironic and fun messages with sarcasm. 18% reflect either supportive ideas or critical politic views.[12] To access more analysis check out Crimson Hexagon’s blog page.[13] The TV Series Kuzey Guney (North South) from Turkey The last example is from the TV series “KuzeyGuney” broadcasted in Turkey. In one episode, one of the characters of the series, Ferhat, was killed, and the murderer was a secret, however, the followers of the series share their ideas about who the murderer was using social media. According to followers, the most possible murderer was at first the character named as Güney. After some more evidence emerged in the next few episodes, the potential murderer was thought to be Sami by the Twitter users in Turkey, and lastly with more clues, followers considered Burak as the murderer. It is interesting that even though it was not yet clear who the murderer was, twitter users had their own opinions on social media. Consequence Web 2.0 technology has become one of the daily technological activities for people globally and also in Turkey. In particular, social media platforms enable users to socially interact online and reflect their opinions on a daily basis. In Turkey, both public and private companies can use the social media as a tool that enables them to get almost instant feedback about their services and products, which users spread what type of ideas, in what scale their goals are realized. Such feedback can enable government organizations and private companies to revise their strategies related to their services and products. References [1] http://oreilly.com/web2/archive/what-isweb-20.html [2] http://www.sde.org.tr/tr/haberler/1775/ sosyal-medyanin-kamu-hizmetleri-vesiyasette-kullanilabilirligi.aspx [3] Jackson, N. A. & Lilleker, D. G.(2009). ‘Building an Architecture of Participation? Political Parties and Web 2.0 in Britain’, Journal of Information Technology & Politics,6:3,232 — 250 [4] agbeat.com—social-networks-nowaccount- for-1-in-5-minutes-of-all-web-use [5] www.newmediatrendwatch.com—87turkey [6] http://mashable.com/2011/06/30/twitter200-million/ [7] http://www.bilisimdergisi.org/pdfindir/ s140/pdf/128-129.pdf [8] www.crimsonhexagon.com [9] http://www.2012presidentialelection news.com/2012-debate-schedule/2012presidential-debate-schedule/ [10] http://usnews.nbcnews.com/_ news/2012/10/24/14674385-social-mediaanalysis-bayonets-fail-to-cut-romney-butoverall-debate-sentiment-swings-obamasway?lite [11] http://www.crimsonhexagon.com/ social-media-analysis-who-shot-ferhat/ [12] http://www.crimsonhexagon.com/ social-media-opinion-fiscal-cliff/ [13] http://www.crimsonhexagon.com/blog/ [14] http://mashable.com/2013/03/09/ twitter-revolution/ [15]http://mashable.com/2013/02/14/ romantic-states-social-media/ [16]http://www.crimsonhexagon.com/ aggregate-opinion-analysis-removeslanguage-barrier/