Twitter平台多模态情感分类数据集

数据集来源于Twitter平台上用户所发的推文，数据采集时间从2022年3月到8月，采集了推特热榜话题下的推文数据。运用了实验室的服务器进行数据采集和处理。为了保证数据质量，从twitter平台采集了270万余条推文数据之后，编写Python脚本对数据按照uuid进行去重，再利用VGG19模型对图片进行分类，机器筛选出有情感意义的图文数据，最后进行人工筛选和标注

This dataset is sourced from tweets posted by users on the Twitter platform. Data collection was conducted between March and August 2022, focusing on tweets under Twitter’s trending topics. All data collection and processing operations were performed using lab servers. To ensure data quality, after initially acquiring over 2.7 million tweets from the Twitter platform, we developed Python scripts to perform UUID-based deduplication. Subsequently, we classified the accompanying images of the tweets using the VGG19 model, filtered out text-image data with emotional significance via automated screening, and completed manual screening and annotation in the end.

搜集汇总

背景与挑战

背景概述

该数据集包含2022年3月至8月Twitter平台上的270万余条推文，经过多模态筛选和人工标注，总数据量为39.47GB，适用于多模态情感分类研究。