Dressipi Collaborates with ACM Recommender Systems (RecSys)

Dressipi Collaborates with ACM Recommender Systems (RecSys)

RecSys is the premier international forum for the presentation of new research results, systems and techniques in the field of product recommendation systems.

Annual challenges are run by RecSys, and in the last few years, leaders in their fields Twitter (2020/21), Trivago (2019), and Spotify (2018) collaborated with RecSys for the challenges specific to their domains.

Dressipi: The Experts in AI-Powered Fashion Retail Personalization

Our focus is to provide the world’s best personalized recommendations and predictions. We do this by taking a domain specific approach across the data we collect and create, how we structure that data and the models we build. Everything we do is optimized to handle the nuances of fashion.

The fashion domain is an interesting one for recommender systems and also relatively new so this year’s RecSys challenge focuses on fashion recommendations. As recognised leaders in fashion AI and the amazing results we get with our fashion recommendation systems (and other solutions), Dressipi have been chosen as the industry partner for the prestigious challenge.

Participants will be responding to the question ‘When given user sessions, purchase data and content data about items, can you accurately predict which fashion item will be bought at the end of the session?’.

Solving the New-User Problem with In-Session Recommendations

It’s important to be able to make recommendations that respond to what the user is doing during the current session to create the best experience possible that results in a purchase. However, nuances in the fashion domain make it notoriously hard to predict:

  • On average 51% of total visitors are new (Dressipi Data) which means there is no historical data available and we solely have to rely on current session activity.

  • Sessions can be pretty short so we need to be able to make accurate predictions as early as possible, before the user bounces.

  • Even for the other half of visitors that have historical data, trends and other external factors change user preferences much more quickly than in other domains, meaning the historical data might no longer be representative of the user’s interests on a case-to-case basis. This places even more importance on having a highly accurate in-session recommender that can be pulled into the mix.

Demonstration of session and purchase data gathered from visitors

We want a recommender that can predict the item the user will purchase as early as possible in the session, but we have to balance that with having more information available for better accuracy of predictions as the session goes on. At some point the recommendation may no longer be useful because the user has worked through a sufficiently long journey and is about to find the item they want themselves, without the intervention of the recommender.

A High Quality & Accurate Dataset

This new-user problem is one of the many issues Dressipi as a fashion-AI specialist is at the forefront of solving so as part of the RecSys challenge, we will be releasing a public dataset of items labeled with content data (such as color, length, neckline, sleeve style, etc.).

Sample of Dressipi product attributes

Fashion-specific item features (content data) are essential to learn as much as possible from only a few user interactions and successfully recommend items. Using the item features, can we be effective at identifying what type of thing the user is looking for and recommend those types of items, after only a few items have been interacted with during the session.

The full dataset will also contain 1.1 million online retail sessions that resulted in at least one item purchased.

Top Recommender Systems Researchers Face Fashion Industry Challenges

This is a great opportunity for Dressipi to engage further with the research community and ensure that we are delivering the best possible recommendations for our customers. It’ll be interesting to see how the top Recommender Systems researchers of the world approach solving the challenges the fashion industry faces.

After the challenge is concluded we are planning to release the dataset and keep it available for researchers to encourage further research and publications in this area.

Check out the full RecSys Challenge 2022 here.


You might also be interested in this article…

Building Fashion Recommendation Systems

Read about building fashion recommendation systems for ecommerce, and how Dressipi overcomes the challenges of personalized product recommendations.