Tough Tables

ToughTables (2T) 数据集是为 SemTab 挑战创建的，总共包括 180 个表。该数据集中的表可以分为两组：控制 (CTRL) 组表和严格 (TOUGH) 组表。 CTRL 组包含 60 个通过查询 DBpedia SPARQL 端点生成的表和从 Wikipedia 收集的表，它们的特点是易于注释。 TOUGH 组包含 120 个表格，主要是从网络上抓取的，其中一些包含拼写错误的单词和昵称/同音词，它们的特点是难以注释。在这两个组中，作者生成了一些表格，他们在收集的表格中添加了噪音。使用 DBpedia (DBP) 类型和实体以及 WikiData (WD) 为两个任务对数据集进行了注释：列类型注释 (CTA) 和单元实体注释 (CEA)。在下表中列出了为 CTA 注释的列数和为 CEA 任务注释的单元格数以及使用的类数。注释类 DBP-列类型注释 540 39 DBP-单元实体注释 663,656 16,023 WD-列类型注释 540 276 WD-单元实体注释 667,244 24,653The ToughTables (2T) dataset was developed for the SemTab Challenge, and it contains a total of 180 tables. The tables in this dataset can be divided into two groups: the Control (CTRL) group and the Tough (TOUGH) group. The CTRL group consists of 60 tables generated by querying the DBpedia SPARQL endpoint and collected from Wikipedia, which are characterized by ease of annotation. The TOUGH group includes 120 tables, most of which are scraped from the web. Some of these tables contain misspelled words, nicknames or homophones, and are characterized by difficulty in annotation. For both groups, the authors generated additional tables by adding noise to the collected tables. The dataset was annotated for two tasks: Column Type Annotation (CTA) and Cell Entity Annotation (CEA), using DBpedia (DBP) types and entities as well as WikiData (WD). The number of annotated columns for CTA, the number of annotated cells for CEA, and the number of utilized classes are listed below: - DBP-Column Type Annotation: 540 columns, 39 classes - DBP-Cell Entity Annotation: 663,656 cells, 16,023 classes - WD-Column Type Annotation: 540 columns, 276 classes - WD-Cell Entity Annotation: 667,244 cells, 24,653 classes

搜集汇总

背景与挑战

背景概述

Tough Tables数据集专为SemTab挑战而构建，包含180个表格，分为易于注释的CTRL组和难以注释的TOUGH组。该数据集支持列类型注释和单元实体注释任务，并使用DBpedia和WikiData进行标注。