Geological Text to Knowledge (GEOTEK) : Extracting Heterogeneous Information from Marine Geology Publications

This thesis presents an innovative approach to automate the information extraction from research publications within the field of Marine Geology. Specifically, the newly developed method targets at extracting core measurements i.e. Mass Accumulation Rate (MAR) and Sedimentation Rate (SR), which are largely reported in relevant publications alongside spatial and contextual information presented in text and tables. This research addresses the challenges faced by existing extraction approaches focusing scientific publications.

Generally, marine geologists rely on manual procedures to extract relevant information and context from scientific publications. However, with the increasing volume of scientific publications, such manual procedures have become tedious and challenging. The widespread use of PDF format for representing scientific publications further instigated information extraction and modelling challenges encompassing diverse data representations and modularity as PDF primarily supports reading rather than automatic information extraction. This study proposes an Information Extraction framework that focuses major data elements of Marine Geology publications. As a building block, it develops a comprehensive Data Acquisition Framework (DAF) capable of efficiently extracting all the data components from scientific publications including metadata, text, tables and figures. Besides, potential applicability across diverse scientific domains. Several information extraction approaches were tailored that extract Oceanographic Named Entities and relevant scientific measurements by targeting plain-text and tables from Marine Geology publications. Furthermore, the parsed information is systematically assembled by Heterogeneous Information Linking (HIL). The compiled information is populated into an interactive map and a novel Marine Geology Document Network, which could enable future knowledge discovery and serve as a gateway to potential applications.

Logo KCSS

Rechte

Nutzung und Vervielfältigung:


CC BY-NC-ND 4.0

Bitte beachten Sie, dass einzelne Bestandteile der Publikation anderweitigen Lizenz- bzw. urheberrechtlichen Bedingungen unterliegen können.

Zitieren

Zitierform:
Suryani, M.A., 2025. Geological Text to Knowledge (GEOTEK): Extracting Heterogeneous Information from Marine Geology Publications. Kiel Computer Science Series. Department of Computer Science, Kiel. https://doi.org/10.21941/kcss/2025/1
Zitierform konnte nicht geladen werden.