I'm trying to research potential ideas for visualizing sentiment analysis - specifically sentiment analysis of Tweets about a subject. I have some data and have bucketed the tweets into three buckets of sentiment (negative, neutral, and positive). Does anyone have any suggestions or examples of visualizing this kind of data? I'm exploring ideas of viewing sentiment over time, but am curious to know if anyone has done any user research or thinking around this topic that would be willing to share their experiences.

  • 1
    Hi Jessica, as it stands now, your question is asking for others to solve your problem. If you provide your ideas for visualizing this, others can critique, suggest changes on it and help you improve it. Look at websites like flowingdata.com/category/visualization nad others for some inspiration.
    – rk.
    Commented May 13, 2013 at 21:07
  • Thanks - was more just looking for some good research starting points and research, than for someone to solve the problem for me.
    – Jessica
    Commented May 13, 2013 at 22:26
  • This How to Chose the Best Chart for Your Data article could be a good starting point. It's chock full of helpful links.
    – Stasome
    Commented May 13, 2013 at 23:33

2 Answers 2


There is an abundance of research on sentiment analysis and within this research you can find several attempts at how to visualize sentiment data. I have found a few sample papers that should get you started [1,2,3]. In these papers systems are presented that use a variety of techniques ranging from the traditional (pie graphs for classification buckets, as in your original question) to more creative (word clouds charted against time, explorable clouds of sentiment groups, sentiment charted on a geographical heatmap, etc.).

The right technique for visualization depends on your usage context. Back in my grad school days I did a project on analyzing the sentiment of tweets about live basketball games - we were able to use simple pie charts and line charts for showing trends in sentiment.

[1] Dongxu Duan, Weihong Qian, Shimei Pan, Lei Shi, and Chuang Lin. 2012. VISA: a visual sentiment analysis system. In Proceedings of the 5th International Symposium on Visual Information Communication and Interaction (VINCI '12). ACM, New York, NY, USA, 22-28. DOI=10.1145/2397696.2397700 http://doi.acm.org/10.1145/2397696.2397700

[2] Amitava Das, Sivaji Bandyopadhyay, and Björn Gambäck. 2012. Sentiment analysis: what is the end user's requirement?. In Proceedings of the 2nd International Conference on Web Intelligence, Mining and Semantics (WIMS '12). ACM, New York, NY, USA, , Article 35 , 10 pages. DOI=10.1145/2254129.2254173 http://doi.acm.org/10.1145/2254129.2254173

[3] Eivind Bjørkelund, Thomas H. Burnett, and Kjetil Nørvåg. 2012. A study of opinion mining and visualization of hotel reviews. In Proceedings of the 14th International Conference on Information Integration and Web-based Applications & Services (IIWAS '12). ACM, New York, NY, USA, 229-238. DOI=10.1145/2428736.2428773 http://doi.acm.org/10.1145/2428736.2428773


Before you do any design, it is always important to look more closely at the type of information that you have, and also the message you want to deliver with the visualization. As you can see from some these examples (here and here), there are some quantitative and qualitative information that you can display in a number of different ways. If your focus is on time, then being able to highlight the changes over specific time period (through either static or interactive means) is going to be a focus, so you need to make sure that the data does reflect some changes or trends rather than try to make the visualization to exaggerate it when there are no specific trends or patterns.

Just remember to also qualify what the sentiment analysis involves, so that the users have a context for the information presented. The rest is just a matter of playing with different design ideas and see what will give you the best way to tell the message you want to provide to the audience.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.