Classifying Indian addresses for the e-commerce domain


Details
Abstract regarding the talk: In the e-commerce industry, where shipments are delivered everyday, understanding addresses is of vital importance to ensure that there are no delays in the shipment. However, in India and several third-world countries, addresses do not follow a prescribed format - a single address can have multiple variants. Parsing such addresses due to lack of inherent structure can be challenging. The talk focuses on this problem and a novel approach of understanding customer addresses in the e-commerce domain by pre-training language models and fine-tuning them for different purposes.
About the speaker: T. Ravindra Babu did M.Sc(Engg) and then Ph.D. from Indian Institute of Science. He currently heads the Data Science team at Sahaj. His previous stints include VP & Head, Data Science team at Myntra, Principal Data Scientist at Flipkart, Principal Researcher at Infosys Research and scientist at ISRO. His interests include machine learning and Spacecraft Orbital Dynamics. He has publications, patents and a book in these areas.

Classifying Indian addresses for the e-commerce domain