Geographic Knowledge Graph Extraction Performance Metrics

2026 data from Zongjun Wei compares the performance of different relation extraction methods for building Geographic Knowledge Graphs. The dataset contains experimental results from a framework tested on the Yangtze River Delta river network. It was published in April 2026.

Use Cases

• Compare F1 scores (0.898 and 0.823) for geographic entity recognition and relation extraction tasks across different methods.
• Benchmark the performance of the LLM-Reflex-GeoKG framework against baseline systems using reported evaluation metrics.
• Analyze the impact of a multi-stage extraction strategy on knowledge graph completeness and accuracy metrics.
• Validate relation extraction techniques using data specific to river networks and watershed management domains.

Strengths

• Includes specific performance metrics (F1 scores of 0.898 and 0.823) for two key NLP tasks.
• Data is recent, published in April 2026.
• Results are from a defined geographic context: the Yangtze River Delta river network.

Limitations

• The dataset is very small at 5.5 KB, indicating limited scope, likely containing only summary results.
• The number of rows, columns, and raw data samples are unknown.
• Geographic coverage is limited to a single river network region.

Provenance

Source

Zongjun Wei via figshare.

Collection Method

Results from experiments validating the LLM-Reflex-GeoKG framework.

Time Range

2026.

Freshness

Published in April 2026.

Geography

Yangtze River Delta river network.