國立高雄大學圖資館 |

Language: English

Back

Efficient Data Systems for Scalable Analysis With Large Language Models /

Record Type:	Electronic resources : Monograph/item
Title/Author:	Efficient Data Systems for Scalable Analysis With Large Language Models /Saehan Jo.
Author:	Jo, Saehan,
Description:	1 electronic resource (224 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 86-12, Section: A.
Notes:	Advisors: Trummer, Immanuel Committee members: Wagner, Aaron; Ellis, Kevin; Kuleshov, Volodymyr.
Contained By:	Dissertations Abstracts International86-12A.
Subject:	Computer science.
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31997347
ISBN:	9798283140506

Efficient Data Systems for Scalable Analysis With Large Language Models /
Jo, Saehan,

Efficient Data Systems for Scalable Analysis With Large Language Models /Saehan Jo. - 1 electronic resource (224 pages)

Source: Dissertations Abstracts International, Volume: 86-12, Section: A.

Modern data analysis increasingly relies on large language models (LLMs) to process diverse data modalities, including text, images, audio, and video. Although LLMs offer advanced reasoning capabilities for unstructured data, their high inference costs significantly exceed those of traditional relational operators. This computational overhead often becomes a critical bottleneck, limiting the scalability of LLM-driven analytical systems in real-world applications.This thesis introduces two complementary systems designed to mitigate this challenge and enable cost-efficient data analysis with LLMs. Both systems leverage principles from approximate query processing to balance computational cost and result quality. The first, ThalamusDB, reduces inference costs by minimizing the volume of data processed. Guided by specialized cost and error models, ThalamusDB identifies an optimal data subset for processing while respecting user-defined error constraints. The second system, SpareLLM, reduces costs by automatically selecting the smallest, most efficient LLM that satisfies user-defined equivalence constraints compared to a powerful reference model. Together, these systems present distinct strategies for achieving scalable data analysis with LLMs. Furthermore, this work introduces BitGourmet and AggChecker to address related challenges in specialized data processing and result verification, respectively.

English

ISBN: 9798283140506Subjects--Topical Terms:

199325
Computer science.
Subjects--Index Terms:

Large language models

Efficient Data Systems for Scalable Analysis With Large Language Models /
LDR:02884nmm a22004213i 4500 001 693694
005 20260511063012.5
006 m o d
007 cr|nu||||||||
008 260819s2025 miu||||||m |||||||eng d
020 $a 9798283140506
035 $a (MiAaPQD)AAI31997347
035 $a AAI31997347
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Jo, Saehan, $e author. $0 (orcid)0009-0001-3880-902X $3 1010320
245 1 0 $a Efficient Data Systems for Scalable Analysis With Large Language Models / $c Saehan Jo.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2025
300 $a 1 electronic resource (224 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 86-12, Section: A.
500 $a Advisors: Trummer, Immanuel Committee members: Wagner, Aaron; Ellis, Kevin; Kuleshov, Volodymyr.
502 $b Ph.D. $c Cornell University $d 2025.
520 $a Modern data analysis increasingly relies on large language models (LLMs) to process diverse data modalities, including text, images, audio, and video. Although LLMs offer advanced reasoning capabilities for unstructured data, their high inference costs significantly exceed those of traditional relational operators. This computational overhead often becomes a critical bottleneck, limiting the scalability of LLM-driven analytical systems in real-world applications.This thesis introduces two complementary systems designed to mitigate this challenge and enable cost-efficient data analysis with LLMs. Both systems leverage principles from approximate query processing to balance computational cost and result quality. The first, ThalamusDB, reduces inference costs by minimizing the volume of data processed. Guided by specialized cost and error models, ThalamusDB identifies an optimal data subset for processing while respecting user-defined error constraints. The second system, SpareLLM, reduces costs by automatically selecting the smallest, most efficient LLM that satisfies user-defined equivalence constraints compared to a powerful reference model. Together, these systems present distinct strategies for achieving scalable data analysis with LLMs. Furthermore, this work introduces BitGourmet and AggChecker to address related challenges in specialized data processing and result verification, respectively.
546 $a English
590 $a School code: 0058
650 4 $a Computer science. $3 199325
650 4 $a Information technology. $3 184390
650 4 $a Information science. $3 190425
653 $a Large language models
653 $a Modern data analysis
653 $a Real-world applications
653 $a Cost-efficient data analysis
690 $a 0984
690 $a 0489
690 $a 0723
710 2 $a Cornell University. $b Computer Science. $3 730438
720 1 $a Trummer, Immanuel $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 86-12A.
790 $a 0058
791 $a Ph.D.
792 $a 2025
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31997347