SemEval 2025 Task 8: Question Answering on Tabular Data

Join the first SemEval task on Question Answering on Tabular Data. Create a system that answers questions related to real-world table datasets and participate in Subtask I or Subtask II. Competition starts on January 10, 2025.

We are excited to announce the first SemEval task on Question Answering on Tabular Data. This task involves creating a system that answers questions related to real-world table datasets from various domains. The datasets have a large number of rows and columns, and a wide variety of data types, allowing for the assessment of different types of questions related to each data type.

Participants will be provided with (dataset, question) pairs and will need to provide an answer which will then be compared with a gold standard. The answer might be achieved through various methods, including In-Context Learning and Code Generation.

There are two subtasks for this competition: Subtask I involves answering questions using the full dataset, while Subtask II involves using a sampled version of each dataset with a maximum of 20 rows per dataset.

Important dates: Competition starts on January 10, 2025 and ends on January 31, 2025.

Tags: SemEval 2025, Question Answering, Tabular Data, DataBench, In-Context Learning, Code Generation, Competition, Subtask I, Subtask II