Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2026
|
| Online Access: | https://doi.org/10.5281/zenodo.20353868 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866902026547363840 |
|---|---|
| author | Ojaswini Nair, Prachi Mahajan, Sandhya Petchimuthu and Meghana Nandala |
| author_facet | Ojaswini Nair, Prachi Mahajan, Sandhya Petchimuthu and Meghana Nandala |
| contents | <p class="MsoNormal">Even small increases in sea surface temperature (SST) are reshaping the distribution of commercially important fish species such as the Indian oil sardine and Indian mackerel, highlighting the growing ecological vulnerability of India’s marine ecosystems under climate variability. Although India possesses extensive oceanographic observations and biodiversity records, much of this information remains fragmented across isolated repositories, limiting its integrated use for ecosystem-level assessment and policy planning. This study presents a prototype Marine Knowledge Graph framework designed to bridge these data silos through semantic integration of heterogeneous marine datasets. The system incorporates real biodiversity and environmental DNA (eDNA) records alongside controlled environmental scenarios developed to evaluate system functionality and scalability. A modular extract–transform–load (ETL) pipeline standardizes oceanographic variables, taxonomic metadata, and molecular sequences before transforming them into an interconnected graph structure linking species, habitats, monitoring nodes, and climate parameters. Species identification is implemented using a k-mer–based genomic similarity approach combined with L2 normalization, UMAP dimensionality reduction, and FAISS nearest-neighbor indexing. Instead of probabilistic classification, detection is performed through embedding-distance similarity matching. Functional validation confirms effective semantic querying, relational linkage, and genomic similarity retrieval within a unified computational environment. The framework provides a scalable foundation for integrative marine biodiversity monitoring and climateresponsive ecosystem management. Keywords: Marine Knowledge Graph, eDNA, k-mer Analysis, UMAP, FAISS, Genomic Similarity Search, Data Integration, Climate Variability, Marine Biodiversity Monitoring</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_20353868 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | A UNIFIED AI-DRIVEN KNOWLEDGE GRAPH PROTOTYPE FOR ENHANCING MARINE ECOLOGICAL RESILIENCE IN THE INDIAN EEZ Ojaswini Nair, Prachi Mahajan, Sandhya Petchimuthu and Meghana Nandala <p class="MsoNormal">Even small increases in sea surface temperature (SST) are reshaping the distribution of commercially important fish species such as the Indian oil sardine and Indian mackerel, highlighting the growing ecological vulnerability of India’s marine ecosystems under climate variability. Although India possesses extensive oceanographic observations and biodiversity records, much of this information remains fragmented across isolated repositories, limiting its integrated use for ecosystem-level assessment and policy planning. This study presents a prototype Marine Knowledge Graph framework designed to bridge these data silos through semantic integration of heterogeneous marine datasets. The system incorporates real biodiversity and environmental DNA (eDNA) records alongside controlled environmental scenarios developed to evaluate system functionality and scalability. A modular extract–transform–load (ETL) pipeline standardizes oceanographic variables, taxonomic metadata, and molecular sequences before transforming them into an interconnected graph structure linking species, habitats, monitoring nodes, and climate parameters. Species identification is implemented using a k-mer–based genomic similarity approach combined with L2 normalization, UMAP dimensionality reduction, and FAISS nearest-neighbor indexing. Instead of probabilistic classification, detection is performed through embedding-distance similarity matching. Functional validation confirms effective semantic querying, relational linkage, and genomic similarity retrieval within a unified computational environment. The framework provides a scalable foundation for integrative marine biodiversity monitoring and climateresponsive ecosystem management. Keywords: Marine Knowledge Graph, eDNA, k-mer Analysis, UMAP, FAISS, Genomic Similarity Search, Data Integration, Climate Variability, Marine Biodiversity Monitoring</p> |
| title | A UNIFIED AI-DRIVEN KNOWLEDGE GRAPH PROTOTYPE FOR ENHANCING MARINE ECOLOGICAL RESILIENCE IN THE INDIAN EEZ |
| url | https://doi.org/10.5281/zenodo.20353868 |