虚假信息知识库

虚假信息知识库为多元异构虚假与不良信息传播理论、模型与知识库的数据库支撑。涵盖网络平台中传播的虚假信息与传播网络中的账号，经过对数据的采集、清洗与分类标注后，得到高质量虚假信息数据，主要的数据类型有以下三种:第一种是社交平台发布的文本信息；第二种是社交平台发布的由文本和图像构成的信息;第三种是社交平台发布的由文本和视频构成的信息。本知识库包含虚假信息库包含人物、时间、地点、议题等要素，包含3020611个实体，3022321条关系，10种虚假信息。账号方面包含123738个较大影响力的账号，账号特征8个。所构建的数据集按照模态分为3个文件夹，图像文件夹下面包含47353个文件，视频文件夹下包含17643个文件，数据量约150GB。The Disinformation Knowledge Base provides database support for theories, models and knowledge bases related to multi-heterogeneous disinformation and harmful information dissemination. It covers disinformation disseminated on online platforms and accounts in dissemination networks. After data collection, cleaning, classification and annotation, high-quality disinformation datasets are obtained. The main data types include the following three categories: 1. Text information published on social platforms; 2. Information composed of both text and images published on social platforms; 3. Information composed of both text and videos published on social platforms. This knowledge base covers elements such as persons, time, locations and topics, and contains 3,020,611 entities, 3,022,321 relationships and 10 types of disinformation. In terms of accounts, it includes 123,738 accounts with considerable influence, along with 8 account characteristics. The constructed dataset is divided into three folders based on modality: the image folder contains 47,353 files, the video folder contains 17,643 files, and the total data volume is approximately 150 GB.

搜集汇总

背景与挑战

背景概述

虚假信息知识库是一个用于支撑多元异构虚假与不良信息传播理论、模型与知识库的数据库，涵盖社交平台中的文本、图像和视频等多模态信息。它包含大量实体、关系和账号数据，数据量约162.54GB，文件格式包括jpg、mp4等。该数据集由北京理工大学创建，属于国家重点研发计划项目的一部分。