Do you like to explore new dishes every time you order food? Or do you stick to your usual favourites? Do you like a Raita with your Biryani, or just a soft drink? Do you prefer veg or non veg? Do these preferences change depending on time of day or day of the week?

These are among the several questions that need to be automatically inferred from data while building a food recommendation engine at scale. Right from ranking dishes on the menu to suggesting complimentary dishes on the cart, Data Science is at the very core of constantly improving our suggestions. In this talk, we will dive into multiple scenarios that pop up while building a food recommendation system in terms of handling data sparsity and handling new restaurants / users among other challenges that crop up due to scale. We will look at different algorithms that need to be explored in order to effectively handle these various challenges.


Outline/Structure of the Talk

Proposed outline:

1. Introduction & Motivation - 7 mins

1.1 Data driven analysis to motivate why it is important to recommend and personalise

1.2 Different scenarios and challenges for food recommendation

1.3 Mathematical formulation of metrics - NDCG, MRR

2. Formulations and algorithms for different scenarios - 10 mins

2.1 Popularity & User Segments

2.2 Learning to Rank Formulation

2.3 Multi arm bandit Formulation

2.4 Multi stakeholder recommender system Formulation

3. Questions - 3 mins

Learning Outcome

* Learning to think about different scenarios, challenges and approaches while building a recommendation engine

* Get a brief introduction towards different learning algorithms like L2R, MAB

Target Audience

Anyone with intermediate understanding of Machine Learning

schedule Submitted 6 months ago

Public Feedback

comment Suggest improvements to the Author
  • Deepti Tomar
    By Deepti Tomar  ~  5 months ago
    reply Reply

    Hello Ashay,

    Thanks for your time and efforts on the proposal! Could you answer the following questions to help the program committee understand your proposal better?

    • Are these demo(s) /use case(s) from your project work? Speaker's experience on the project helps people understand the concept better.
    • Are you also planning to share the results of how you improved the food recommendation system by using the techniques mentioned? 



    • Ashay Tamhane
      By Ashay Tamhane  ~  5 months ago
      reply Reply

      Thanks Deepti for considering the proposal. Yes, the use cases are from my current project at Swiggy. I will be sharing the results as well.

      • Deepti Tomar
        By Deepti Tomar  ~  5 months ago
        reply Reply

        Thanks for your response, Ashay! We will let you know in case if we have more questions.