Welcome to the NTCIR-19 DAGRI pilot task, a challenge in analyzing agricultural documents containing complex diagrams and tables, aiming to a new framework for knowledge sharing in local agriculture, laying the foundation for nurturing future farmers and enhancing the sustainability of the agricultural sector.

DAGRI consists both a subtask of digitalization to convert some complex data in the agricultural documents into machine-readable formats and a subtask of understanding domain-specific knowledge by building a question-answering system from these documents which are inherently difficult for machines to interpret directly.

1. Table IE
   This subtask aims to extract agricultural management types and indicators from various representations of tabular data contained in PDF files, and convert that information into unified format.

2. Multi-Table Multi-Hop QA
   This subtask aims to answer each of the question about farming operations precisely, given the multiple tables labelled "Technical Systems" in agricultural documents.

3. FiT-QA
   This subtask aims to develop methods for answering questions based on figures and tables containing complex layouts, implicit structures and domain-specific terminology that are inherently difficult for machines to interpret directly.

• January, 2026: Distribution of Example Dataset

• February, 2026: 1st Orientation Meeting of DAGRI

• March, 2026: Distribution of Preliminary Test Dataset

• April, 2026: 2nd Orientation Meeting of DAGRI

• March – June, 2026: Preliminary Test (Dry Run)

• July, 2026: Formal Test (Formal Run)

• August 1, 2026: Return of Evaluation Results to Participants, Partial release of task overview paper

• September 1, 2026: Deadline for Draft Papers from Task Participants

• November 1, 2026: Final Submission Deadline for Task Papers and Task Overview Papers

• December 8–10, 2026: NTCIR-19 Conference (NII, Tokyo)