Article Summarization

Article Summarization / Text Summarization

Necessity and Importance

With the ever-increasing volume of information on the web and the significant growth in published articles across various scientific fields, quick access to and study of required information has always been a challenge for researchers.

Therefore, the Noor Computer Research Center for Islamic Sciences (Noor), which consistently advances research facilitation through its products, has launched the Noor Text Summarization System by employing expert personnel and utilizing the latest software capabilities.

System Capabilities

Using this system, you can:
- Find and categorize important sections of any text
- Summarize an article or research paper
- Easily study the summaries

Algorithm and Implementation Method

In developing the Noor Text Summarization System, center specialists focused on methods that could display rich, fluent, and smooth summaries in the shortest possible time.

Therefore, the Text Rank algorithm (for detecting and identifying the most important sentences in a text) was used in implementing this system. This algorithm can summarize lengthy texts with very high speed and in minimal time with excellent accuracy and quality.

Quality Evaluation

The summarization method used in this system, in addition to initial evaluations, has also been examined using the ROUGE evaluation method (which compares machine-generated summaries with human-generated summaries).

Furthermore, to ensure maximum output quality, center engineers compared summaries generated by this system with abstracts written by article authors. Ultimately, the system's output demonstrated a higher confidence level.

Practical Application

This system is currently being utilized in the Noor Specialized Journals Database (Noormags) to produce summaries for articles in this database, and the results are available as "Machine Summary".

Benefits

• High Speed: Summarizing lengthy texts in minimal time
• Desired Quality: Producing rich, fluent, and smooth summaries
• Acceptable Accuracy: Validated by ROUGE evaluation method and comparison with author abstracts
• Practical Application: Successfully deployed on Noormags platform