SERAG: Semantic Entity Retrieval from Arabic knowledge Graphs

This is the dataset used in: "S. Esmeir, SERAG: Semantic Entity Retrieval from Arabic knowledge Graphs, In: Proceedings of the Sixth Arabic Natural Language Processing Workshop (WANLP 2021) at EACL 2021". The dataset is a translation of a subset (139/467) of the queries in DBpedia Entity v2 to Modern Standard Arabic. We used the “stopped” version of DBpedia Entity v2 (queries-v2_stopped.txt). Please use the query ID to link the translated queries to their English counterparts, and to the relevance judgment files provided with DBpedia Entity v2 (qrels-v2.txt). DBpedia’s interlingual mapping file (interlanguage_links_ar.ttl.bz2) can be used to map entities from English to Arabic and vice-versa. Please refer to the paper for translation considerations and baseline results.