ConsBASE: End-to-end ETL pipeline for Korean construction document integration

Seung Kyu Hong; Dong Seok Seo; Bum Soo Kim; Hyuk Yoon Kwon

doi:10.1016/j.softx.2026.102696

ConsBASE: End-to-end ETL pipeline for Korean construction document integration

Seung Kyu Hong
, Dong Seok Seo
, Bum Soo Kim
, Hyuk Yoon Kwon

Dept. of Industrial Engineering/ITM

Seoul National University of Science and Technology (SNUST)
Korean Institute of Civil Engineering and Building Technology

Research output: Contribution to journal › Article › peer-review

Abstract

Construction documents are predominantly stored as unsearchable scanned images, hindering automated data utilization. We introduce ConsBASE , an automated ETL (Extract, Transform, Load) pipeline tailored for Korean construction documents. ConsBASE extracts unstructured content by recognizing complex tables and text from raw images, normalizes the extracted features into a unified JSON schema, and loads the structured data into a searchable database. Accompanied by a web-based user interface, ConsBASE eliminates manual document inspection, bridging the gap between static archives and active data utilization. This software enables advanced analytics and significantly enhances data accessibility in construction workflows.

Original language	English
Article number	102696
Journal	SoftwareX
Volume	34
DOIs	https://doi.org/10.1016/j.softx.2026.102696
State	Published - Jun 2026

Keywords

Construction CALS
Digitalization
Figure data
Optical character recognition
Table data

Access to Document

10.1016/j.softx.2026.102696

Cite this

@article{5c7e412e28b54d13b21647b90b567c7d,

title = "ConsBASE: End-to-end ETL pipeline for Korean construction document integration",

abstract = "Construction documents are predominantly stored as unsearchable scanned images, hindering automated data utilization. We introduce ConsBASE , an automated ETL (Extract, Transform, Load) pipeline tailored for Korean construction documents. ConsBASE extracts unstructured content by recognizing complex tables and text from raw images, normalizes the extracted features into a unified JSON schema, and loads the structured data into a searchable database. Accompanied by a web-based user interface, ConsBASE eliminates manual document inspection, bridging the gap between static archives and active data utilization. This software enables advanced analytics and significantly enhances data accessibility in construction workflows.",

keywords = "Construction CALS, Digitalization, Figure data, Optical character recognition, Table data",

author = "Hong, \{Seung Kyu\} and Seo, \{Dong Seok\} and Kim, \{Bum Soo\} and Kwon, \{Hyuk Yoon\}",

note = "Publisher Copyright: {\textcopyright} 2026 The Author(s).",

year = "2026",

month = jun,

doi = "10.1016/j.softx.2026.102696",

language = "English",

volume = "34",

journal = "SoftwareX",

issn = "2352-7110",

}

TY - JOUR

T1 - ConsBASE

T2 - End-to-end ETL pipeline for Korean construction document integration

AU - Hong, Seung Kyu

AU - Seo, Dong Seok

AU - Kim, Bum Soo

AU - Kwon, Hyuk Yoon

PY - 2026/6

Y1 - 2026/6

N2 - Construction documents are predominantly stored as unsearchable scanned images, hindering automated data utilization. We introduce ConsBASE , an automated ETL (Extract, Transform, Load) pipeline tailored for Korean construction documents. ConsBASE extracts unstructured content by recognizing complex tables and text from raw images, normalizes the extracted features into a unified JSON schema, and loads the structured data into a searchable database. Accompanied by a web-based user interface, ConsBASE eliminates manual document inspection, bridging the gap between static archives and active data utilization. This software enables advanced analytics and significantly enhances data accessibility in construction workflows.

AB - Construction documents are predominantly stored as unsearchable scanned images, hindering automated data utilization. We introduce ConsBASE , an automated ETL (Extract, Transform, Load) pipeline tailored for Korean construction documents. ConsBASE extracts unstructured content by recognizing complex tables and text from raw images, normalizes the extracted features into a unified JSON schema, and loads the structured data into a searchable database. Accompanied by a web-based user interface, ConsBASE eliminates manual document inspection, bridging the gap between static archives and active data utilization. This software enables advanced analytics and significantly enhances data accessibility in construction workflows.

KW - Construction CALS

KW - Digitalization

KW - Figure data

KW - Optical character recognition

KW - Table data

UR - https://www.scopus.com/pages/publications/105038545188

U2 - 10.1016/j.softx.2026.102696

DO - 10.1016/j.softx.2026.102696

M3 - Article

AN - SCOPUS:105038545188

SN - 2352-7110

VL - 34

JO - SoftwareX

JF - SoftwareX

M1 - 102696

ER -

ConsBASE: End-to-end ETL pipeline for Korean construction document integration

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this