Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387235533> ?p ?o ?g. }
Showing items 1 to 63 of
63
with 100 items per page.
- W4387235533 abstract "Abstract The oil and gas (O&G) industry generates a significant volume of data from various sources, such as seismic surveys, well logs, drilling reports and others, which are stored in either relational or non-relational databases. However, searching for pertinent data records requires the end-user to be conversant with database query syntaxes and schema definitions, which is challenging. In this paper, we introduce a novel framework to interact with O&G databases using natural language. We trained a Text-to-Text Transfer Transformer (T5) (Raffel, 2020) in a multi-task setting incorporating natural language to structured query language (Text-to-SQL) conversion as a main task with query context classification, and passage context classification as auxiliary tasks. We also introduce a method for data augmentation with SQL to natural language task (SQL-to-Text). Additionally, we implemented database-aware query disambiguation for typos and spelling correction by incorporating string and phonetics similarity algorithms. We evaluated the performance of our trained language model (LM) on a test dataset consisting of 1711 natural language queries involving diverse O&G entities, such as fields, wellbore, well logs, markers, etc. Our innovative multi-task approach resulted in a remarkable performance, achieving an exact set match accuracy (EM) of 88.5% on the Text-to-SQL task with impressive F1 scores of 96.3% for query context classification, and 87.7% for passage context classification tasks. We observed multi-task training of LM improved the EM accuracy of the SQL predictions. We also demonstrate the application of our approach to enable contextual natural language search on Open Subsurface Data Universe (OSDUTM) platform. Overall, our approach has potential to transform how end-users search and retrieve data from O&G databases." @default.
- W4387235533 created "2023-10-02" @default.
- W4387235533 creator A5023997949 @default.
- W4387235533 creator A5033227639 @default.
- W4387235533 creator A5038921095 @default.
- W4387235533 date "2023-10-02" @default.
- W4387235533 modified "2023-10-16" @default.
- W4387235533 title "Enabling Contextual Natural Language Search on Oil and Gas Databases" @default.
- W4387235533 cites W2945102109 @default.
- W4387235533 cites W3124687886 @default.
- W4387235533 cites W3214600982 @default.
- W4387235533 cites W4307259665 @default.
- W4387235533 cites W4385573058 @default.
- W4387235533 doi "https://doi.org/10.2118/216349-ms" @default.
- W4387235533 hasPublicationYear "2023" @default.
- W4387235533 type Work @default.
- W4387235533 citedByCount "0" @default.
- W4387235533 crossrefType "proceedings-article" @default.
- W4387235533 hasAuthorship W4387235533A5023997949 @default.
- W4387235533 hasAuthorship W4387235533A5033227639 @default.
- W4387235533 hasAuthorship W4387235533A5038921095 @default.
- W4387235533 hasConcept C154945302 @default.
- W4387235533 hasConcept C164120249 @default.
- W4387235533 hasConcept C174252522 @default.
- W4387235533 hasConcept C194222762 @default.
- W4387235533 hasConcept C195324797 @default.
- W4387235533 hasConcept C204321447 @default.
- W4387235533 hasConcept C23123220 @default.
- W4387235533 hasConcept C41008148 @default.
- W4387235533 hasConcept C510870499 @default.
- W4387235533 hasConcept C55596503 @default.
- W4387235533 hasConcept C5655090 @default.
- W4387235533 hasConcept C77088390 @default.
- W4387235533 hasConcept C97854310 @default.
- W4387235533 hasConceptScore W4387235533C154945302 @default.
- W4387235533 hasConceptScore W4387235533C164120249 @default.
- W4387235533 hasConceptScore W4387235533C174252522 @default.
- W4387235533 hasConceptScore W4387235533C194222762 @default.
- W4387235533 hasConceptScore W4387235533C195324797 @default.
- W4387235533 hasConceptScore W4387235533C204321447 @default.
- W4387235533 hasConceptScore W4387235533C23123220 @default.
- W4387235533 hasConceptScore W4387235533C41008148 @default.
- W4387235533 hasConceptScore W4387235533C510870499 @default.
- W4387235533 hasConceptScore W4387235533C55596503 @default.
- W4387235533 hasConceptScore W4387235533C5655090 @default.
- W4387235533 hasConceptScore W4387235533C77088390 @default.
- W4387235533 hasConceptScore W4387235533C97854310 @default.
- W4387235533 hasLocation W43872355331 @default.
- W4387235533 hasOpenAccess W4387235533 @default.
- W4387235533 hasPrimaryLocation W43872355331 @default.
- W4387235533 hasRelatedWork W2183574112 @default.
- W4387235533 hasRelatedWork W2288984523 @default.
- W4387235533 hasRelatedWork W2554350299 @default.
- W4387235533 hasRelatedWork W2624697218 @default.
- W4387235533 hasRelatedWork W2803426644 @default.
- W4387235533 hasRelatedWork W2966628213 @default.
- W4387235533 hasRelatedWork W3192936206 @default.
- W4387235533 hasRelatedWork W4231027902 @default.
- W4387235533 hasRelatedWork W4285324809 @default.
- W4387235533 hasRelatedWork W4312419558 @default.
- W4387235533 isParatext "false" @default.
- W4387235533 isRetracted "false" @default.
- W4387235533 workType "article" @default.