多源异质图建模与业务需求主动预测数据集

本数据集来自于Douban和Yelp等多个第三方服务领域，共包含Movie、Book和Business三个服务领域的用户历史交互数据和各类关系数据，其中Movie数据集包含用户、电影、群组、演员、导演和题材等关系数据；Book数据集包含用户、书籍、作者、出版商和年份等关系数据；Business数据集包含用户、商业、偏好、分类和城市等关系数据，数据集的学科范围属于个性化推荐和需求建模。为了保证数据集的质量，本数据集从原始数据中移除评分较低的交互数据，并将显式反馈转换为隐式反馈。处理后的完整数据集的各类数据共计1,752,595条，文件数据量合计19.3MB，包含的数据格式为.txt。

This dataset is collected from multiple third-party service platforms including Douban and Yelp. It covers three service domains: Movie, Book, and Business, and contains user historical interaction data and various relational data across these domains.
Specifically, the Movie dataset includes relational data related to users, movies, groups, actors, directors, and genres; the Book dataset encompasses relational data involving users, books, authors, publishers, and publication years; and the Business dataset holds relational data about users, businesses, preferences, categories, and cities.
The academic scope of this dataset falls within the fields of personalized recommendation and requirements modeling.
To ensure data quality, low-rated interaction entries were removed from the original dataset, and explicit feedback was converted into implicit feedback. The processed complete dataset totals 1,752,595 records across all categories, with a combined file size of 19.3 MB, and all data is stored in .txt format.

搜集汇总

背景与挑战

背景概述

该数据集整合了来自豆瓣、Yelp等多个平台在电影、书籍和商业领域的用户交互与关系数据，经过去除低评分和反馈类型转换处理，共包含超过175万条数据。它主要用于支持个性化推荐、需求建模及异质信息网络相关的分析与预测研究。