Class Exercise – Flickr Thought Experiment

This week’s NM3229 lecture was pretty action packed as we had a guest lecture by Mr David Ayman Shamma, a senior research scientist at Yahoo! Research in USA (his bio can be found at Bio). Mr Shamma gave a very interesting lecture on what methods are used to analyze and visualize photographs uploaded to social media websites such as Flickr. He demonstrated some innovative techniques to analyze the photos uploaded by various communities on Flickr and also how the data related to the photos could be visualized.

The class exercise for the lecture was to think of a novel way to visualize the data associated with 1.2 millions photos taken in Singapore. For each photo, the following information was available:

  1. Location
  2. Time at which photo is taken
  3. Time at which photo is uploaded to Flickr
  4. User who uploaded the photo

The class split into groups, each group having around 3-4 members. All groups were given 20 minutes to think of a motivating question they would like to answer with their visualization and also to make a rough sketch of the visualization.

My group decided that taking pictures of food is very popular in Singapore. Moreover, getting the location where food pictures are taken and at what time of day they are taken could depict which are some of the most popular eating places in Singapore and at what times of the day are they most crowded. The motivating question we decided to answer with our visualization was – What are the food habits of Singapoeans?

Our rough sketch of the visualization looked as follows:


We conveyed the following information with this interactive visualization:

  1. Locations in Singapore where food photos have been taken. These are represented using a marker in a map of Singapore.
  2. Slider to depict time of day – You can slide to a particular time of day and the photos taken at that time of day will appear at their corresponding locations. Colour codes on markers are used to indicate different times of day. For example, black marker indicates midnight.
  3. User track – by clicking on an individual marker, a connecting line between that photo and other photos uploaded by the same user appears helping you track where in Singapore a user is eating. It could provide interesting information about the eating habits of Singaporeans.

Overall, Mr Shamma and the class were appreciative of our efforts. Mr Shamma mentioned that our visualization was comprehensive and conveyed sufficient meaningful information. The only recommendation he had was to limit the time slider to blocks of hours in a day (eg: breakfast time – 7 am to 11 am , lunch time – 11.30 am to 2 pm etc) instead of for every hour in the day.

The other groups in class also made interesting visualizations. Here is a list of their motivating questions + visualizations:

1. What are the popular hotspots visited in Singapore from 2004 – 2014?


2. What is the difference between time a photo is taken and time it is uploaded to Flickr?


3. What are good places in Singapore to take light photos?


4. What are the most popular colours photographed in Singapore?


5. What do Singaporeans do over the weekend?


Overall, it was a really fun lecture and I was able to learn more about the analysis and visualization of data from photographs through the lecture and class exercise.