An animated picture says at least a thousand words: Selecting Gif-based Replies in Multimodal Dialog

TL;DR

This paper introduces a large dataset and a novel multimodal model for selecting gif-based replies in online conversations, improving the relevance and reception of animated responses in dialogue systems.

Contribution

It presents a new dataset of 1.56 million text-gif conversation turns and a multimodal model, Pepe the King Prawn, for selecting appropriate gif responses in dialogue.

Findings

Model produces relevant and high-quality gif responses.

Gifs generated by the model are significantly better received by users.

The dataset enables training of multimodal conversational models.

Abstract

Online conversations include more than just text. Increasingly, image-based responses such as memes and animated gifs serve as culturally recognized and often humorous responses in conversation. However, while NLP has broadened to multimodal models, conversational dialog systems have largely focused only on generating text replies. Here, we introduce a new dataset of 1.56M text-gif conversation turns and introduce a new multimodal conversational model Pepe the King Prawn for selecting gif-based replies. We demonstrate that our model produces relevant and high-quality gif responses and, in a large randomized control trial of multiple models replying to real users, we show that our model replies with gifs that are significantly better received by the community.