Spider4SSC - A semantically equivalent text-to-SQL/SPARQL/Cypher dataset

Spider4SSC is a composite and extension of Spider (Yu et al., 2018), Spider4SPARQL (Kosten et. al., 2023) and our own Spider4Cypher data. Spider4SSC is a multi-domain text-to-query language dataset with 4525 unique (question, sql, sparql, cypher) samples where semantically matching queries are provided in SQL, SPARQL and Cypher. The dataset contains 159 distinct domain relational databases from Spider \cite{yu-etal-2018-spider} and 159 equivalent graph databases. This allows cross-query-language benchmarking and fine-tunning of neural networks on text-to-query task.