Designing the Future of Personal Fashion


Consumers and industry are highly interested in the potential of personalized fashion. In this work, we both develop models to automatically capture the elements of fashion style and develop tools to expose consumer pain points and opportunities for tech interventions in fashion.

Our Approach

PSBot Chatbot Architecture

In our first exploration in this space, we develop a model that can learn latent fashion concepts from data. Using polylingual topic models (PLTMs), we learn latent fashion concepts joinly in two languages: a style language describing outfits and an element language labeling clothing items. This model allows us to translate between the two languages, exposing the elements of fashion style.

The second project investigates consumer pain points and opportunities for tech interventions. To do so, we conduct two need-finding studies that explore the gold-standard of personalized shopping: interacting with a personal stylist. These results capture both interviews with working personal stylists as well as conversational logs from user interactions with a personal stylist chatbot -- PSBot.

Code and Data

Fashion data (from Polyvore) used to train the PLTM can be found here.
Code for the PSBot chatbot can be found here.


Kristen Vaccaro, Tanvi Agarwalla, Sunaya Shivakumar and Ranjitha Kumar. Designing the Future of Personal Fashion. CHI 2018. pdf

Ranjitha Kumar and Kristen Vaccaro. An Experimentation Engine for Data-Driven Fashion Systems. AAAI 2017 Spring Symposium. pdf

Kristen Vaccaro, Sunaya Shivakumar, Ziqiao Ding, Karrie Karahalios and Ranjitha Kumar. The Elements of Fashion Style. UIST 2016. pdf


Kristen Vaccaro

Ranjitha Kumar