The Problem: By default, Elasticsearch treats Urdu/Arabic like English. It fails to recognize Right-to-Left (RTL) logic, character normalization, or diacritics (Harakat).
For a specialized library like the Sirah Research Hub, Elasticsearch is far superior to the legacy Zebra engine—but only if you install the "Missing Piece."
1. The Missing Piece: The ICU Analysis Plugin
The ICU (International Components for Unicode) plugin is vital. It allows the search engine to ignore Zer/Zabar/Pesh so a user finds the Prophet's name regardless of exact spelling.
# Run this on your Koha server terminal:
sudo /usr/share/elasticsearch/bin/elasticsearch-plugin install analysis-icu
sudo systemctl restart elasticsearch
2. Configure the Arabic/Urdu Analyzer
Once the plugin is active, tell Koha to use it. Navigate to:
Koha Admin → Search Engine Configuration (Elasticsearch)
Ensure your multilingual fields are mapped to the icu_analyzer or the dedicated arabic analyzer to handle word roots and stemming.
Zebra vs. Elasticsearch: The Sirah Hub Verdict
| Feature | Zebra (Legacy) | Elasticsearch (Recommended) |
|---|---|---|
| Urdu/Arabic Support | Requires manual .chr file hacking | Native with ICU Plugin |
| Performance | Slow with 14,000+ records | Sub-millisecond response |
| Fuzzy Searching | Basic/Limited | Advanced (Finds "Sirah" vs "Seerah") |
| Scalability | Fixed/Rigid | Built for Big Data |
💡 Academic Impact: Elasticsearch’s "Fuzzy Searching" is a lifesaver for researchers. It ensures that varying transliterations of names (e.g., Shibli vs Shebli) still lead to the correct scholarly records.
Action Step: Verify with your technical team if analysis-icu is installed. Without it, the Hub is limited; with it, it is world-class.
