Under the coding hood: InOut 6.0
October 19th, 20th 2019 at MLR Convention Centre, Bengaluru
Please note: Pranav, a good junior of mine has a notable write up as well. Please do give that a read before this one:
The first time I heard about InOut Hackathon was in 2018, when Adarsh was shortlisted to attend it. From twitter, I learned how cool the event was and wished to attend it in the future.
Fast forward about 10 months later, portals are open yet again. I applied solo and (unsuprisingly) was kept in waiting list. My hopes were not too high as I knew where I stood relative to the top coders of my institution and so knew odds were not quite in my favor.
However, as by some good act of kindness from a greater force, I was selected as another individual canceled their invitation. On one hand, I was quite contended but quickly experienced impostor syndromes and feared being called a fraud in tech over there. I did not have any field of expertise and tend to spend more taking breadth-first exploration style instead of depth-first mastery.
The very same Adarsh convinced me to apply and be his teammate and I started to feel very much better. And so, the journey began.
Kurian (my classmate), Pranav (my junior from college), Naseem (good friend from CUSAT) and myself boarded the Bangalore express train from Ernakulam Town station on Friday evening. We spoke of a lot of nerdy technical matters on the way.
We reached Majestic railway station next day morning at 7.30 AM and boarded the metro to J P Nagar — about an hour long metro ride. With another auto rickshaw ride, we made it to the venue and checked in!
Aswin G, Aswin M and Joyal (juniors from college) also made it there on time. Adarsh S came about half an hour later in his expectedly-rather-chill fashion.
He explained the idea as follows:
We need an interface to input the name of an organization and output whether or not they have negatively impacted the Eco-system in the past.
We shall explore each phase in sufficient depth to understand how we can provide this risk assessment technique — interleaving with our experience.
We had a conference in the morning of 19th October.
I skipped a few sessions to interact with other folks in the room. We had participants from different parts of the nation — Noida, Delhi, Kolkata etc. Also the competition was quite high at the scale of IIT’s and IIIT’s. This meant a lot of scope to learn from others during the hackathon!
We approached Roshan Alexander, a fellow Malayali who was working at Pratilipi. He had taken a talk explaining the tech stack and workflow within the company that helped them scale to cater to the needs of several thousands of users. It was interesting to learn his journey from Munnar Engineering College to several exploratory years in Bombay leading him to the Technical Architect position in Pratilipi.
I saw someone where an Uchiha Clan hoodie and that naturally became the conversation starter for me. Gaurav was from Bangalore and he loves hardware hacks very much. We spoke in depth about Anime, technology and everything else in between.
In the afternoon, we all went to Pizza Hut nearby and had a good time hanging out together.
Hackathon began at 5.30 and that was when our idea began to gain clarity.
Phase 1: Web Crawling
We need to find local news articles about the organization. We target local news outlets because these sources are likely to catch on the issue earlier on than the major mainstream ones. Through this process, we find several hundreds of reports that might be relevant to our problem.
Phase 2: Sentiment Analysis
From these articles, we perform a sentiment analysis to understand the underlying tone of the article. We then filter out the positive articles and focus solely on the negative or neutral articles as these are the ones that are of more interest to our problem.
Phase 3: Named Entity Recognition
After selecting the right articles to consider, we carry out some NLP techniques — in particular, Named Entity Recognition to understand the type of terms used in the sentences. A good NER distinguishes ‘apple’ in ‘Apple earned $1 Billion’ as an organization, while ‘apple’ in ‘Apples from market’ as a fruit. This will help us understand the key terms to consider for our problem.
Phase 4: Knowledge Graph Generation
This is the essential portion of our work where we convert words into visual equivalent. An example of this case — eg: Apple earned $1 Billion.
Thus we create a graph database through the knowledge graph generation process, thus taking us a step closer to solving the problem.
Phase 5: Graph Query Validation
Using the graph database created thus far, we check to see if there are any relationships within the system that we do not approve of which harm the ecosystem.
The following resources help us better understand the technical challenges involved at each stage, and how to overcome it:
Apart from the hackathon project, there sure were a lot of things happening:
- Awesome goodies and stickers.
- Fun filled game events at 3 AM (type racing).
- Food every couple of hours.
I went to sleep at 3 AM. Thankfully mattresses , pillows and blankets were provided so had a comfortable sleep till 7 AM. Regarding our project, Adarsh did most of the work while I tried to learn from bits and pieces online and tried to implement it.
An interesting new fact that I learned from another participant who was working was that Java is favored over Python due to clarity provided in syntax (variable type is specified so other developers will have lesser confusion). He encouraged me to continue on the path of exploring tech and said it was never too late. Grateful for his kind words.
For lunch, my sister, cousin brother, and his sisters fiance took me and Adarsh out. We had a good conversation and appreciate the extreneous effort they all took to meet up, considering how long travel takes in Bengaluru.
Adarash and Ashuthosh (new member who joined us. Studies at IIT Bombay) presented the idea as the rest of us had to leave to catch the train back home. I am really grateful to Adarsh for nurturing me during the hackathon and ensuring I learned a thing or two.
On the train ride back home:
- Interacted with an entrepreneur who was trying out Influencer Marketing strategies on YouTube. He explained the perks of doing an MBA and we discussed other aspects of life
- Interacted with students from Law College. Watching them be well versed in the IPC’s and opening us to nuances in cases and possible errors made me appreciate the work advocates do.
We reached Ernakulam 9 AM next day morning (1.5 hours late). Pranav, Naseem and Kurian came over to my place for breakfast.
After an engaging game of chess with Pranav, they departed and the journey truly concluded.
Overall, it was a great learning experience and this was definitely the biggest hackathon I have been to in the recent past.
