Myntra is one of India’s leading fashion e-commerce companies, delivering a best-in-class shopping experience through advanced machine learning models. This session will delve into a key machine learning solution designed to enhance query understanding for product search flow. Our ML model accurately interprets user intent from all types of search queries, helping shoppers find exactly what they are looking for without sifting through irrelevant results. We will discuss various iterations and enhancements of this solution.
Improving the understanding of search queries by developing a multilabel text classification model that maps search queries to possible shopping intents.
Example:
Search query: “winter upper wear”
Intents: jackets, sweaters
It has two key components:
- Data and Feature Engineering: Myntra constructs robust training datasets by leveraging user query reformulation and detailed product catalog attributes that encompass a wide range of potential search terms and user intents.
- Neural Network Modeling: Our neural classifier, powered by text embeddings and a custom neural network, excels in accurately understanding shopping intents.
Introduction to the problem: 5 minutes
Data and feature engineering: 7.5 minutes
Neural network modeling: 7.5 minutes
Explainability within model parameters: 5 minutes
Conclusion and Q&A: 5 minutes
This solution has significantly reduced query abandonment rates, increased click-through rates, and boosted overall user engagement and revenue.
Who is this talk for?
This session is ideal for Data/Applied Scientists, Data Engineers, and ML Engineers who face challenges in creating well-annotated supervised training data from user interactions on their platforms. It’s particularly relevant for professionals working on scalable products involving information retrieval, regardless of their level of experience.
Presenter : https://www.linkedin.com/in/pushkar-aggrawal-739361175/
{{ gettext('Login to leave a comment') }}
{{ gettext('Post a comment…') }}{{ errorMsg }}
{{ gettext('No comments posted yet') }}