BriefGPT.xyz
Jul, 2022
AutoWeird: 随机搜索发现的奇怪翻译评分函数
AutoWeird: Weird Translational Scoring Function Identified by Random Search
HTML
PDF
Hansi Yang, Yongqi Zhang, Quanming Yao
TL;DR
本篇论文描述了一种名为AutoWeird的奇怪评分函数，其只使用三元组中的尾实体和关系来计算其可信度得分，并且在ogbl-wikikg2数据集上取得了排名第一的性能，这可能源于不恰当的评估和集中的尾实体分布。
Abstract
scoring function
(SF) measures the plausibility of triplets in knowledge graphs. Different scoring functions can lead to huge differences in
link prediction
performances on different knowledge graphs. In this rep
→