Osman Readability Metric

📢 Now Available in textstat

The Osman Readability Metric for Arabic text is now integrated into the textstat Python package.

📦 Installation

pip install textstat

▶️ Example Usage

import textstat

text = "أدخل النص العربي هنا"
score = textstat.osman(text)
print(score)

This returns the Osman readability score for your Arabic input text.

---

🧠 What is Osman?

Osman is a readability metric specifically designed for Arabic, inspired by traditional metrics like Flesch and Fog, but enriched with a unique component called Faseeh—a linguistic factor that accounts for Arabic fluency and complexity.

Originally developed in Java, the Osman tool supports texts with or without diacritics (Tashkeel). When Tashkeel is absent, the tool estimates syllables automatically for readability scoring.

📄 Reference:
El-Haj, M., and Rayson. "OSMAN - A Novel Arabic Readability Metric".
LREC 2016, Portoroz, Slovenia.
Paper PDF | Lancaster Link

---

🚀 What's New in 2020?

• Major code update on 14 August 2020
• No longer depends on Farasa due to licensing
• Fully open-source and free
• Switched to lightweight RegEx-based tokenisation (instead of Stanford)

These changes reduce storage requirements and improve runtime speed.

---

🖥️ How to Run the Java Tool

You can use the graphical tool via the pre-built JAR file.

▶️ Run with GUI:

1. Download Osman2020.jar from the releases.
2. In your terminal or command prompt, run:

java -Dfile.encoding=UTF-8 -jar Osman2020.jar

💡 If you encounter Arabic display issues on Windows, you can:

• Change your system locale to Arabic (see YouTube Tutorial)
• Run this first: set JAVA_TOOL_OPTIONS=-Dfile.encoding=UTF-8

---

🔧 Java Source Code

Developers can clone the repo and use the classes directly:

• Use OsmanReadability.calculateOsman(String text)
• Strip diacritics using removeTashkeel(String text)
• Example usage: See TestOSMAN.java

The tool also calculates other readability scores such as ARI and LIX.
📌 Ensure UTF-8 encoding is set for console output in your IDE.

---

🧰 Import into Eclipse (Beginner Friendly)

1. Install EGit via Help → Install New Software
2. Add the Juno update site, search for git, install EGit + Mylyn GitHub connector
3. Go to File → Import → Git → Projects from GitHub
4. Search OsmanReadability and select drelhaj/OsmanReadability
5. Finish setup and explore the project

---

📚 OSMAN UN Corpus

• Find data under the Osman_UN_Corpus directory
• To download: click on a zip file → View Raw to start download
• Or use the Download ZIP button on the GitHub repo

---

📬 Contact

For questions or feedback, email:
📧 dr.melhaj@gmail.com