# Automatic Fashion Knowledge Extraction from Social Media

## TL;DR

This paper introduces a system that automatically extracts comprehensive fashion knowledge from social media by integrating multiple tasks and modalities, leveraging contextual learning and weak label modeling to improve accuracy.

## Contribution

It presents a novel unified framework for occasion, person, and clothing discovery from images, texts, and metadata, with innovative use of contextualized learning and noise reduction techniques.

## Key findings

- Improved fashion concept learning performance.
- Effective handling of noisy training data.
- Demonstration website showcasing high-quality knowledge extraction.

## Abstract

Fashion knowledge plays a pivotal role in helping people in their dressing. In this paper, we present a novel system to automatically harvest fashion knowledge from social media. It unifies three tasks of occasion, person and clothing discovery from multiple modalities of images, texts and metadata. A contextualized fashion concept learning model is applied to leverage the rich contextual information for improving the fashion concept learning performance. At the same time, to counter the label noise within training data, we employ a weak label modeling method to further boost the performance. We build a website to demonstrate the quality of fashion knowledge extracted by our system.