An Ontology of Well Engineering Entities to Extract and Structure Text Data from Daily Reports

Registro completo de metadados
MetadadosDescriçãoIdioma
Autor(es): dc.contributorUniversidade Estadual Paulista (UNESP)-
Autor(es): dc.contributorUniversidade Estadual de Campinas (UNICAMP)-
Autor(es): dc.contributorPetrobras-
Autor(es): dc.creatorQueiroz, J.-
Autor(es): dc.creatorJaculli, M. A.-
Autor(es): dc.creatorJunior, N. Choueri-
Autor(es): dc.creatorSilveira, I. M.-
Autor(es): dc.creatorMendes, J. R.P.-
Autor(es): dc.creatorPenteado, B. E.-
Autor(es): dc.creatorGuilherme, I. R.-
Autor(es): dc.creatorPerrout, S. R.-
Data de aceite: dc.date.accessioned2025-08-21T23:09:40Z-
Data de disponibilização: dc.date.available2025-08-21T23:09:40Z-
Data de envio: dc.date.issued2025-04-29-
Data de envio: dc.date.issued2024-12-31-
Fonte completa do material: dc.identifierhttp://dx.doi.org/10.2118/223801-MS-
Fonte completa do material: dc.identifierhttps://hdl.handle.net/11449/305003-
Fonte: dc.identifier.urihttp://educapes.capes.gov.br/handle/11449/305003-
Descrição: dc.descriptionOil and gas activities generate, inevitably, huge amounts of data, especially in the form of Daily Operational Reports during construction and production activities. Extracting information from these reports can be achieved using embeddings combined with Large Language Models (LLMs), with domain knowledge being mandatory for this extracted information to be meaningful and properly presented to end-users. A solution for this challenge is the use of Knowledge Graphs (KG) to organize and relate all the information in a way that properly represents the concepts (e.g., entities, events, and processes) of the specific domain. A KG can address the lack of domain knowledge presented by LLMs, as well as enhance the semantics of embedding-based indexes such as those provided by Transformers, but also as a searching source itself that enables to take advantage of the information relationships (i.e., the natural links between entities or pieces of data). However, the creation of a KG from a specific domain based on unstructured data sources requires not only an ontology to describe the entities and their relationships but also strategies and tools capable of properly extracting and mapping them from the data sources to a KG. In this context, this work proposes a well engineering ontology to guide extracting information from Daily Operational Reports, structure them into KGs, and support information retrieval applications. To assure interoperability, the development of the proposed ontology is aligned with upper-level ontologies and industrial standards like ISO 15926. The results of this work include not only the ontology itself and the procedure followed to build it, but also a discussion of further applications that can be developed using it as a basis. The main applications resulting from this work are named entity recognition, visualization of incidents, classification of operations, templates and auto-filling of reports, calculation of drilling metrics, and queries within the reports database. We have noticed that having structured domain knowledge, in the form of an ontology and KG has improved the success of these applications compared to raw text processing. Finally, this work emphasizes the importance of domain knowledge in the form of a comprehensive ontology that focuses on the relevant entities of the application domain and enables the creation of KGs that can facilitate the organization and query of complex unstructured text information and enhance data analysis and LLM-based applications in the Oil&Gas domain.-
Descrição: dc.descriptionUniversidade Estadual Paulista-
Descrição: dc.descriptionSão Paulo State University, São Paulo-
Descrição: dc.descriptionUniversity of Campinas, São Paulo-
Descrição: dc.descriptionPetrobras, Rio de Janeiro-
Descrição: dc.descriptionSão Paulo State University, São Paulo-
Descrição: dc.descriptionUniversidade Estadual Paulista: 2019/00697-8-
Idioma: dc.languageen-
Relação: dc.relationSPE - International Association of Drilling Contractors Drilling Conference Proceedings-
???dc.source???: dc.sourceScopus-
Título: dc.titleAn Ontology of Well Engineering Entities to Extract and Structure Text Data from Daily Reports-
Tipo de arquivo: dc.typeaula digital-
Aparece nas coleções:Repositório Institucional - Unesp

Não existem arquivos associados a este item.