Machine learning & NLP at Goibibo: Keytopic extraction and contextual sentiment analysis of 20 million reviews, and a customer assistance bot

Dhruv Pathak (~dhruv58)


26

Votes

Description:

  • We, at Goibibo, extensively use Python as part of our app as well as data processing stack.
  • We receive a good volume of user generated content, and use natural language processing and machine learning to derive insights and data from them for our products.
  • The talk would cover how we build our user review analyzer system which extracts keytopics/keywords from 20 million user reviews and user sentiments about that keytopic. For example : How we detect that "the rooms were very spacious" is a positive experience for a consumer, or that a particular hotel has an amazing candle light dinner.

  • The talk would cover language processing in Python using inbuilt/installed python modules, and dedicated NLP libraries like NLTK and spaCy, and sentiment analysis/categorization done by Scikit-learn.

This feature can be viewed and experienced on Goibibo mobile apps ( Android/ iOS) and visiting a hotel's product page.

enter image description here

Prerequisites:

  • Intermediate level knowledge of Python is beneficial, however the concepts are language agnostic .
  • Knowledge of NLTK , spaCy and scikit-learn is a plus.

Content URLs:

Talk presentation link : https://prezi.com/p/coenajjzcuis/

Speaker Info:

I am Dhruv Pathak, a python enthusiast. I currently work at Goibibo as VP of Engineering, where we ship many evolving and high scale products using Python.

Speaker Links:

  • https://stackoverflow.com/users/533399/dhruvpathak
  • https://www.linkedin.com/in/dhruvpathak
  • https://github.com/dhruvpathak

Section: Others
Type: Talks
Target Audience: Intermediate
Last Updated:

Hi,

Please upload the talk presentation for review by the Pycon India team.

Thanks!

Gourav Chawla (~Gouravchawla)

Added in the proposal. Also pasting here for reference : https://prezi.com/view/m2Fe65t1RTuQK2s1LZrC/

Dhruv Pathak (~dhruv58)

Login to add a new comment.