Question-Answer matching task is one of the most important tasks in Natural Language Processing(NLP). It has many real world applications in domains like customer support, human resources, etc. Most of the Question-Answer matching systems depend on explicit feature-engineering and are thus domain-specific. An end-to-end system with no or little feature-engineering is therefore highly desirable. Here, we present one such system implemented using techniques of Deep Learning. We show that such a system requires no featuring engineering and yet is powerful enough to attain competitive performance when compared to other state-of-the-art models.
Bio: Bhanu is working as a Senior Data Scientist @ Talla Inc., Cambridge MA. He studied Informatik at University of Bonn, where he worked on Neural Embeddings and Recursive Neural Networks for NLP. There after, he has been actively involved in implementing solutions for various real-world problems using Machine Learning and NLP. He can be contacted at: firstname.lastname@example.org.
Resources: Presentation slides