Sugandh Sinha


Contribution

Data analysis and preparation

Since, data was the core to getting started with this project, We started researching various data formats, available at KI mesh site, to see which one would be more efficent. We found 'tree.bin' which had nodes and its child separated by a '.' as a delimiter. For instance, the node C26.986.950 hierarchy can be broken down as C26 (root) -> 986 -> 950 and is translated to Wounds and Injuries -> Wounds, Penetrating -> Wounds, Stab.

The task was to convert nodes of the form C26.986.950 into Wounds and Injuries.Wounds, Penetrating.Wounds, Stab. For this, I wrote a Python program which used regex. The major problem that I faced was doing a parallel lookup for the updated nodes. I also tried to convert the csv file into json format but the hierarchy created problems for csv to json conversion.

Although, later, we found a niche solution (found by Andreas) to use data in JSON format directly from KI mesh database server. Total time spent : 9 - 10 hours.

Visualization design and testing

Defined the main outline for the design of visualization with other members and how the mechanisms should work. I also tested the visualizations at various stages and gave feedback. Total hours spent : 5-6 hours

Setting up a full stack server - major fail!

I tried to set up a full stack server using Ubuntu server with mongodb for database, flask as web framework and python, html so that the search could be fast but after setting it up, I realized that although it would return the result as JSON object for the searched term, making it work for clicking on individual node might end up taking a lot of time and definitely, not the efficient or the easiest way around, especially considering the case that it my first time setting up a full stack server! Total time spent : 8 - 9 hours.

Visualizations

In the beginning, all of us were trying to come up with our own visualizations. I created two such visualizations using the csv formatted data file that I created. First, I created radial dendrogram (all nodes expanded) using only the top 1500 elements from the csv file.

Then I created a dendrogram based visualizations and another tabular visualization with all nodes present but it was taking a lot of time to load.

Later, for the the final visualization, I also did a few bug fixes like making the placement of the root nodes properly aligned when you load the page and along with Lili, polishing of the final visualization like making the search node a bit eye-catcy, etc. Total time spent : 10 hours

Video preparation

Also, I made the video for the final submission to the C-awards. Total time spent : 8-9 hours.

Meetings

I attended almost all the meetings where we defined goals and discussed the problems/issues. Also, having a meeting with the KI people before we started anything, helped a lot! Total time spemt : 8-9 hours

What I have learned

Well, I think defining proper goals for individuals or groups is important and then following it up is equally important. We had a lot of communication problems and thus, everything was going haywire at one point in time when everyone was working at their own code. Also, what I noticed is that I don't think we followed agile methodology for development, it could have helped somewhat with everybody working on the same thing.

But, it felt really nice to know that this project can be potentially used by a lot of students.