Date: 24th June, 2017
Hey all,
We are all set for the 7th session of our meetup. Couple of months back we had Krupal Modi with us from Haptik.ai discussing on how to build chatbots and the process behind it, starting from rule based system to more ML based approaches. This time we go beyond rule based systems to dig deeper into ML based techniques with Aditya Joshi, Senior Data Scientist in the Cuddle.ai.
Abstract of the Session:
Natural Language Processing (NLP) deals with computer programs that process natural language text for specific tasks. A more-than-half-a-century-old area of research, NLP has been traditionally viewed as an expert system with a set of human-defined rules. However, with textual data proliferating on the web, NLP systems have long moved on to statistical implementations that use large-size textual corpora.
In this talk, Aditya will share his experiences in three research areas of statistical NLP: machine translation, opinion mining, and sarcasm detection. Machine translation aims to translate sentences from a source language to a target language. He will describe the use of Expectation-Maximization (EM) algorithm to find translations of words from parallel sentences in the two languages. The second area is opinion mining: detection of opinion in text. In this case, he will focus on the use of Latent Dirichlet Allocation (LDA) model to discover political issues and orientations from a largely unsupervised corpus. The third area is sarcasm detection where he will discuss the use of Hidden Markov Models (HMMs) for sarcasm detection of conversations from a TV series 'Friends'.
About Speaker:
Aditya Joshi is currently a Senior Data Scientist in the Cuddle.ai. He completed his MTech in Computer Science and Engineering from IIT Bombay in 2011, and is nearing the completion of his PhD at IITB-Monash Research Academy, a dual PhD programme between IIT Bombay and Monash University, Australia. Aditya’s research, primarily in natural language processing, spans innovation (drunk-texting prediction, political orientation prediction, measuring sentiment complexity by recording eye movements, etc.) and fundamental studies (sarcasm detection, word senses for sentiment analysis, etc.). He has published in leading NLP forums around the world and has conducted training sessions in NLP for diverse audiences.
4:00 PM - Introduction 4:15 PM - Isn't NLP all about a bunch of rules? by Aditya Joshi 5:30 PM - Tea.
Uploaded in the folder.
- Book for Statistical NLP mentioned by Aditya - Foundations of Statistical NLP