Towards becoming a Full Stack Computational Linguist
Recently, I had the great opportunity to attend an advanced summer school at Language Technologies Research Centre (LTRC), IIIT Hyderabad. To provide a background, the best work in the domain of NLP in India happens at two centres — CFILT, IIT Bombay and LTRC, IIIT Hyderabad. LTRC works very closely with the industry on cutting edge NLP problems and advancing research into Indian Languages.
The summer school is sponsored by IIIT Hyderabad and Google. It is a two week school on Natural Language Processing with lectures and project guidance by internationally renowned researchers in the area. The summer school is project oriented, where each of the participants choose a project from a list in advance. During the summer school the participants work, under the supervision of experts in the respective area, to come up with substantial advancement in their chosen project. Participants then present their project at the end of the school.
STARTING THE NLP SUMMER SCHOOL
The summer school was an intense bootcamp, with lectures from 9.30 AM — 1 PM. The lectures covered areas like;
- Machine Translation
- Parsing
- Morphological Analysis
- Sentiment Analysis
- Question Answering System
- From context to meaning: Application of Distributional semantics
- Information Extraction and Information Retrieval
- Speech Recognition
This was followed by project work in a hackathon environment, where work used to extend till midnight. My project involved creating a framework for a recommendation engine, using user browser history and user-generated reviews using word2vec. We made something that not only was able to mimic the existing recommender systems deployed by websites, but goes one step ahead — incorporating context to generate valid and innovative recommendations. The beauty of such a framework is that not only does it support online learning, but is also sensitive to minor changes in user tone and behavior.
[caption id=”attachment_151" align=”alignnone” width=”698"]
Late night architectures for word2vec recommendation systems[/caption]
I would like to thank Dr. Manish Shrivastava, Mr. Vandan, Akhil Gupta and Vinay Agrawal on providing crucial inputs and making the project successful.Overall, it was a brilliant experience, and the academic setting helped focus my energies on the tasks at hand. No office deadlines, client emails or worrying about live projects. It was such a bliss !
I would want my network to benefit from my experience too. So, I am going to publish my notes from the lectures to help budding NLP practitioners over the next few weeks. The notes would cover basics of linguistics, speech recognition, parsing, Machine translation, QA sytems and sentiment analysis. I would greatly welcome any feedback, hoping that it this minor contribution helps permeate NLP across the Indian industry, which is not only scalable, but also has firm footing in academic principles.