Enregistré dans:
Détails bibliographiques
Auteurs principaux: Wang, Shuting, Liu, Yunqi, Yang, Zixin, Hu, Ning, Dou, Zhicheng, Xiong, Chenyan
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2506.01689
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866909632806518784
author Wang, Shuting
Liu, Yunqi
Yang, Zixin
Hu, Ning
Dou, Zhicheng
Xiong, Chenyan
author_facet Wang, Shuting
Liu, Yunqi
Yang, Zixin
Hu, Ning
Dou, Zhicheng
Xiong, Chenyan
contents Querying generative AI models, e.g., large language models (LLMs), has become a prevalent method for information acquisition. However, existing query-answer datasets primarily focus on textual responses, making it challenging to address complex user queries that require visual demonstrations or explanations for better understanding. To bridge this gap, we construct a benchmark, RealVideoQuest, designed to evaluate the abilities of text-to-video (T2V) models in answering real-world, visually grounded queries. It identifies 7.5K real user queries with video response intents from Chatbot-Arena and builds 4.5K high-quality query-video pairs through a multistage video retrieval and refinement process. We further develop a multi-angle evaluation system to assess the quality of generated video answers. Experiments indicate that current T2V models struggle with effectively addressing real user queries, pointing to key challenges and future research opportunities in multimodal AI.
format Preprint
id arxiv_https___arxiv_org_abs_2506_01689
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Respond Beyond Language: A Benchmark for Video Generation in Response to Realistic User Intents
Wang, Shuting
Liu, Yunqi
Yang, Zixin
Hu, Ning
Dou, Zhicheng
Xiong, Chenyan
Artificial Intelligence
Computation and Language
Querying generative AI models, e.g., large language models (LLMs), has become a prevalent method for information acquisition. However, existing query-answer datasets primarily focus on textual responses, making it challenging to address complex user queries that require visual demonstrations or explanations for better understanding. To bridge this gap, we construct a benchmark, RealVideoQuest, designed to evaluate the abilities of text-to-video (T2V) models in answering real-world, visually grounded queries. It identifies 7.5K real user queries with video response intents from Chatbot-Arena and builds 4.5K high-quality query-video pairs through a multistage video retrieval and refinement process. We further develop a multi-angle evaluation system to assess the quality of generated video answers. Experiments indicate that current T2V models struggle with effectively addressing real user queries, pointing to key challenges and future research opportunities in multimodal AI.
title Respond Beyond Language: A Benchmark for Video Generation in Response to Realistic User Intents
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2506.01689