Draw Me Like My Triples - 3IA Côte d’Azur – Interdisciplinary Institute for Artificial Intelligence Access content directly
Conference Papers Year : 2023

Draw Me Like My Triples

Abstract

Humans are critical for the creation and maintenance of high-quality Knowledge Graphs (KGs). However, creating and maintaining large KGs only with humans does not scale, especially for contributions based on multimedia (e.g. images) that are hard to find and reuse on the Web and expensive to generate by humans from scratch. Therefore, we leverage generative AI for the task of creating images for Wikidata items that do not have them. Our approach uses knowledge contained in Wikidata triples of items describing fictional characters and uses the fine-tuned T5 model based on the WDV dataset to generate natural text descriptions of items about fictional characters with missing images. We use those natural text descriptions as prompts for a transformer-based text-to-image model, Stable Diffusion v2.1, to generate plausible candidate images for Wikidata image completion. We design and implement quantitative and qualitative approaches to evaluate the plausibility of our methods, which include conducting a survey to assess the quality of the generated images.
Fichier principal
Vignette du fichier
Draw_Me_Like_My_Triples___Wikidata_Workshop_CameraReady.pdf (2.03 Mo) Télécharger le fichier
Origin : Publisher files allowed on an open archive

Dates and versions

hal-04262826 , version 1 (27-10-2023)

Identifiers

  • HAL Id : hal-04262826 , version 1

Cite

Raia Abu Ahmad, Martin Critelli, Şefika Efeoğlu, Eleonora Mancini, Celian Ringwald, et al.. Draw Me Like My Triples: Leveraging Generative AI for Wikidata Image Completion. Wikidata 2023 - The 4th Wikidata Workshop (at ISWC 2023), Nov 2023, Athenes, Greece. ⟨hal-04262826⟩
52 View
25 Download

Share

Gmail Facebook X LinkedIn More