Building slackbots, upping the NLP game & side-project ideas | Weekly Report 40

15 May 2017 . category: report . Comments
#kaggle #data science #slack #nlp

My weekly report from the 8th of May to the 14th of May – A highly productive week at work, not so much on kaggle. Tradeoffs!

Continuing on my efforts to automate Customer Success jobs, i’ve built a slackbot inspired by SalesMachine. Started on Monday, got the Version 1.0 up and running by Friday! Experimented with the RealTime and the Events API, and then decided to stick with slash commands, funny i know ;) The bot serves up basic metrics, which often require either CSMs or analysts time and effort. Still a work in progress, as i have to integrate Celery or Redis to make it asynchronous, probable demo this week :)

On the Quora challenge, i’ve been making pretty slow progress – my meagre ram has been holding back my submissions. I’ve exhausted almost all of my NLP ideas – implemented chunking, NER, chunk hashing, and tried clustering on named entities(failed yet again!). Now, i’m moving on to word embeddings features involving word2vec, learning about it! Apparently, there are some leaks which the top rankers have figured out – mind-bending!

I’ve been thinking of taking up a serious side-project for a while now, went through a lot of ideas lately. James Altucher would be proud ;) Style Transfer in cricket commentary(text), Voice style transfer(inverse dubsmash!), an AR idea, tweet suggestions, anonymous support groups – Just ideas though :(

Data side projects are in very nascent stages – Still ideating on the “mainstream media bias” project, which narrative to build on, what to explore etc. Currently in the process of requesting data from Pilani for our “Placement Analysis” – looking unlikely. The Fedal blog got good recognition, valuable feedback from sports journalists and a feature in an Australian sports analytics blog too. Feedback loops!

On the fitness front, went for a couple short runs on the weekend, and a couple of intense workouts. Not a lot though, gotta focus! No learning, no reading – Bad.

Anyways. Later :)

---