WellQuest

Published: 3 July 2026| Version 1 | DOI: 10.17632/f64ny7mcvz.1
Contributors:
,
,
,
,

Description

WellQuest is an annotated question-answering dataset derived from Daily Drilling Reports (DDRs) in petroleum well engineering. The dataset includes 390 Portuguese question-answer pairs from two offshore drilling interventions, as well as anonymized DDR corpora for both wells. Each item contains a question, a short answer, an expanded answer, linked DDR row identifiers, contiguous activity context, and classification labels across seven dimensions. To broaden accessibility, an English translation of the free-text fields is provided (WellQuest_v1.0_EN.xlsx), produced by LLM-assisted machine translation under an expert-reviewed, domain-specific glossary; the original Portuguese workbook (WellQuest_v1.0.xlsx) remains the authoritative version. A bilingual Portuguese–English glossary of drilling terminology is also included (glossary_PT_EN.csv). WellQuest is designed to support evaluation and development of information retrieval, question answering, and retrieval-augmented generation systems for semi-structured technical documents in petroleum engineering, including cross-lingual and English-language settings.

Files

Institutions

Categories

Information Retrieval, Natural Language Processing, Well Engineering, Natural Language Generation, Deepwater Drilling

Licence