Tags

, , ,

Social Media Monitoring

The proliferation of social media networks today provides organizations with valuable information with regards to how they are being discussed: what is said and what is the view of their clients about them [2]. This kind of information should help an organization improve its services or use the statistics for various purposes like making predictions [1]. Thus, social media monitoring enables an organization to keep an eye on different social websites; deciding what exactly to monitor and which keywords to use to find the content is up to each company to determine according to its social media strategy.
Numerous tools exist for such tasks; Radian6  and Crimson Hexagon are a couple of examples. Also Addicto-matic  which simply displays the latest updates or trends from the different social media websites like Twitter, Flickr, and Youtube.

Twitter Visualization Tools

For Twitter specifically, there exists several tools online that allow us to visualize what is happening with respect to a certain term or user. One can for example view a network showing the relationships between people. One can view on a map where tweets originate; plus any number of statistics imaginable can be computed like the number of tweets per day, per topic etc. One example is TwitterStreamGraphs which displays a simple stream (usage over time) for a given keyword. While TwitterMap displays tweets on a map. Other tools can be found here.

Analysis of the network

Since we are speaking of a social network, we may be interested in the characteristics exhibited by the network. Therefore, we may study the network itself. This provides insights for different domains in terms of better understanding the opportunities available and their impact [3,4]. For instance, using the following ratio: ratio of a user’s followers to the number of people he’s following, the “influence” a user has may be quantified [3].

Analysis of the content

However, in addition or instead of the structure of the network, we may also be interested in the content provided by each user. Many number of characteristics can be examined on twitter. Retweetability- the probability that a tweet will be retweeted by other users- has been examined in [5] where it was found that tweets with URLs are more likely to be retweeted. Sentiment analysis, discussed next, is another aspect.

Sentiment Analysis Tools

Sentiment Analysis, aka. opinion mining, deals with identifying emotions from text using natural language processing tools. Again, there are many online tools that perform sentiment analysis on tweets. Examples include tweetfeel.com which lets you type in some keywords; the tweets are then classified as either “positive” or “negative”, and marked correspondingly in green or red. Another is twitrratr.com which distinguishes between “positive”, “neutral”, and “negative”. Also socialmention.com which actually searches for content in many sites like Youtube and Flickr. And in addition to the “positive”, “neutral”, and “negative” classification, it provides other quantities like “strength”: the likelihood that the term is being discussed, and “passion”: the likelihood that the term will be discussed again by those talking about it. More tools are listed here.
It should be noted that it is very difficult to detect sarcasm which leads to wrong results (a negative being interpreted as a positive). Moreover, most of the tools deal exclusively with the English language; and it is not clear how well the methods used would perform if applied on a different language.

Media Control Rooms

Finally, in an organization that wants to perform social media monitoring and use tools similar to the ones discussed prior, a media control room is often employed. Generally, the control room features a number of screens to monitor different aspects of interest according to the organization; typically a number of visualizations is displayed to the operators there who make use of it in different ways.
A real example of a control room is one used by Gatorade, a brand of sports drinks. They feature six screens showing a number of visualizations like tweets about their brand, other competitive brands, and commercials, showing a difference between more important and less important tweets. Visualizations for blog conversations are also included. More interestingly, they run a sentiment analysis.
Another example is the SocialSphere (by iStrategyLabs) for the Autostadt which displays data gathered in real time from Twitter, Foursquare, and Flickr. A number of applications including a heat map of Foursquare, tweets displayed on top of Flickr images, geocoded tweets, and also a word cloud is visualized. The system is available online here.

References

1. A Framework for Summarizing and Analyzing Twitter Feeds. Xintian Yang, Amol Ghoting, Yiye Ruan, Srinivasan Parthasarathy. KDD 2012.
2. Twitter Sentiment Analysis: The Good, the Bad, and the OMG!. Efthymios Kouloumpis, Theresa Wilson, Johanna Moore. Proceedings of the 5th International AAAI Conference on Weblogs and Social Media.
3. Twitter: Network Properties Analysis. Abraham Ronel Mart nez Teutle. IEEE 2010.
4. Measurement and Analysis of Online Social Networks. Alan Mislove, Massimiliano Marcon, Krishna P. Gummadi, Peter Druschel, Bobby Bhattacharjee. ICM 2007.
5. Want to be Retweeted? Large Scale Analytics on Factors Impacting Retweet in Twitter Network. Bongwon Suh, Lichan Hong, Peter Pirolli, Ed H. Chi. IEEE International Conference on Social Computing 2010.

Advertisements