Matches in SemOpenAlex for { <https://semopenalex.org/work/W4366731740> ?p ?o ?g. }
Showing items 1 to 85 of
85
with 100 items per page.
- W4366731740 abstract "A long standing goal of the data management community is to develop general, automated systems that ingest semi-structured documents and output queryable tables without human effort or domain specific customization. Given the sheer variety of potential documents, state-of-the art systems make simplifying assumptions and use domain specific training. In this work, we ask whether we can maintain generality by using large language models (LLMs). LLMs, which are pretrained on broad data, can perform diverse downstream tasks simply conditioned on natural language task descriptions. We propose and evaluate EVAPORATE, a simple, prototype system powered by LLMs. We identify two fundamentally different strategies for implementing this system: prompt the LLM to directly extract values from documents or prompt the LLM to synthesize code that performs the extraction. Our evaluations show a cost-quality tradeoff between these two approaches. Code synthesis is cheap, but far less accurate than directly processing each document with the LLM. To improve quality while maintaining low cost, we propose an extended code synthesis implementation, EVAPORATE-CODE+, which achieves better quality than direct extraction. Our key insight is to generate many candidate functions and ensemble their extractions using weak supervision. EVAPORATE-CODE+ not only outperforms the state-of-the art systems, but does so using a sublinear pass over the documents with the LLM. This equates to a 110x reduction in the number of tokens the LLM needs to process, averaged across 16 real-world evaluation settings of 10k documents each." @default.
- W4366731740 created "2023-04-24" @default.
- W4366731740 creator A5011478764 @default.
- W4366731740 creator A5039749428 @default.
- W4366731740 creator A5070267604 @default.
- W4366731740 creator A5075032133 @default.
- W4366731740 creator A5084812990 @default.
- W4366731740 creator A5087259526 @default.
- W4366731740 creator A5091512433 @default.
- W4366731740 date "2023-04-19" @default.
- W4366731740 modified "2023-10-17" @default.
- W4366731740 title "Language Models Enable Simple Systems for Generating Structured Views of Heterogeneous Data Lakes" @default.
- W4366731740 doi "https://doi.org/10.48550/arxiv.2304.09433" @default.
- W4366731740 hasPublicationYear "2023" @default.
- W4366731740 type Work @default.
- W4366731740 citedByCount "0" @default.
- W4366731740 crossrefType "posted-content" @default.
- W4366731740 hasAuthorship W4366731740A5011478764 @default.
- W4366731740 hasAuthorship W4366731740A5039749428 @default.
- W4366731740 hasAuthorship W4366731740A5070267604 @default.
- W4366731740 hasAuthorship W4366731740A5075032133 @default.
- W4366731740 hasAuthorship W4366731740A5084812990 @default.
- W4366731740 hasAuthorship W4366731740A5087259526 @default.
- W4366731740 hasAuthorship W4366731740A5091512433 @default.
- W4366731740 hasBestOaLocation W43667317401 @default.
- W4366731740 hasConcept C111472728 @default.
- W4366731740 hasConcept C115903868 @default.
- W4366731740 hasConcept C127413603 @default.
- W4366731740 hasConcept C134306372 @default.
- W4366731740 hasConcept C136197465 @default.
- W4366731740 hasConcept C136764020 @default.
- W4366731740 hasConcept C138885662 @default.
- W4366731740 hasConcept C154945302 @default.
- W4366731740 hasConcept C15744967 @default.
- W4366731740 hasConcept C177264268 @default.
- W4366731740 hasConcept C183003079 @default.
- W4366731740 hasConcept C199360897 @default.
- W4366731740 hasConcept C201995342 @default.
- W4366731740 hasConcept C2522767166 @default.
- W4366731740 hasConcept C2776760102 @default.
- W4366731740 hasConcept C2779530757 @default.
- W4366731740 hasConcept C2780451532 @default.
- W4366731740 hasConcept C2780767217 @default.
- W4366731740 hasConcept C33923547 @default.
- W4366731740 hasConcept C36503486 @default.
- W4366731740 hasConcept C41008148 @default.
- W4366731740 hasConcept C542102704 @default.
- W4366731740 hasConceptScore W4366731740C111472728 @default.
- W4366731740 hasConceptScore W4366731740C115903868 @default.
- W4366731740 hasConceptScore W4366731740C127413603 @default.
- W4366731740 hasConceptScore W4366731740C134306372 @default.
- W4366731740 hasConceptScore W4366731740C136197465 @default.
- W4366731740 hasConceptScore W4366731740C136764020 @default.
- W4366731740 hasConceptScore W4366731740C138885662 @default.
- W4366731740 hasConceptScore W4366731740C154945302 @default.
- W4366731740 hasConceptScore W4366731740C15744967 @default.
- W4366731740 hasConceptScore W4366731740C177264268 @default.
- W4366731740 hasConceptScore W4366731740C183003079 @default.
- W4366731740 hasConceptScore W4366731740C199360897 @default.
- W4366731740 hasConceptScore W4366731740C201995342 @default.
- W4366731740 hasConceptScore W4366731740C2522767166 @default.
- W4366731740 hasConceptScore W4366731740C2776760102 @default.
- W4366731740 hasConceptScore W4366731740C2779530757 @default.
- W4366731740 hasConceptScore W4366731740C2780451532 @default.
- W4366731740 hasConceptScore W4366731740C2780767217 @default.
- W4366731740 hasConceptScore W4366731740C33923547 @default.
- W4366731740 hasConceptScore W4366731740C36503486 @default.
- W4366731740 hasConceptScore W4366731740C41008148 @default.
- W4366731740 hasConceptScore W4366731740C542102704 @default.
- W4366731740 hasLocation W43667317401 @default.
- W4366731740 hasOpenAccess W4366731740 @default.
- W4366731740 hasPrimaryLocation W43667317401 @default.
- W4366731740 hasRelatedWork W1581382276 @default.
- W4366731740 hasRelatedWork W1987766735 @default.
- W4366731740 hasRelatedWork W2026901334 @default.
- W4366731740 hasRelatedWork W2108595774 @default.
- W4366731740 hasRelatedWork W2160548044 @default.
- W4366731740 hasRelatedWork W2374116733 @default.
- W4366731740 hasRelatedWork W2404948224 @default.
- W4366731740 hasRelatedWork W2892218161 @default.
- W4366731740 hasRelatedWork W4294862481 @default.
- W4366731740 hasRelatedWork W4312548019 @default.
- W4366731740 isParatext "false" @default.
- W4366731740 isRetracted "false" @default.
- W4366731740 workType "article" @default.