Text Classification

Noor Text Classification System

The Noor Text Classification System can operate in both Arabic and Persian languages. The use of machine learning has given this classifier acceptable flexibility in dealing with unseen texts and words. The numerous applications of text categories and classes in various sciences and technologies highlight the importance of this system.

What is Text Classification?

Text classification refers to the process of assigning one or more topics or labels to a text based on its content. This process is considered a fundamental operation in natural language processing and text mining. Unstructured texts are abundant, and classifying them is a necessity.

The process of identifying a text's topic can serve as the infrastructure for other natural language processing tasks such as machine translation, optical character recognition (OCR), and speech-to-text conversion.

Text classification has a wide range of applications, from spam detection to sentiment analysis.

Importance of Scientific Text Classification

Classifying scientific texts makes them more accessible to researchers in various fields, especially when the topics are diverse and align with recognized categories.

The Main Challenge

The main challenge in using text categories is that published texts in cyberspace and other scientific (seminary and academic) spaces may either lack any classification or not be classified according to a researcher's perspective.

The computational solution to this challenge is the use of a text classification system.

Noor Text Classification System

At Noor Center, text classification has been developed in two different domains, though other domains can also be addressed based on user needs:

1. Persian News Texts
2. Arabic Jurisprudence (Fiqh) Texts

Persian News Text Classification

Currently, Persian news text classification is performed in two modes: 7-class and 10-class.

7-Class Classification includes:
- Economic
- Social
- Accidents/Incidents
- Foreign
- Political
- Technology
- Sports

10-Class Classification includes:
- Literature & Art
- Short News
- Stock Exchange & Banking
- Global Economy
- Social
- Science & Culture
- Economic
- Tourism
- Miscellaneous
- Accidents/Incidents

For example, a text can be classified as an economic text based on the 10-class Persian classification.

Arabic Jurisprudence (Fiqh) Text Classification

This classification is performed on Arabic texts and can classify Arabic Fiqh texts into the following nine classes:

• Al-Qisas (Retaliation)
• Al-Diyat (Blood Money)
• Al-Hajj (Pilgrimage)
• Al-Makasib (Earnings)
• Al-Mirath (Inheritance)
• Al-Nikah (Marriage)
• Al-Salah (Prayer)
• Al-Taharah (Purity)
• Al-Zakah (Alms)

Development Capability

The Noor Text Classification System has been developed based on the current and potential needs of respected researchers. Therefore, through greater and better interaction between this center and researchers and scholars in the field of Islamic and human sciences, more diverse classifiers in other domains can be achieved.

Since the structure of the machine learning-based text classifier is language-independent and independent of language structure, new classifiers—even in other languages—can be created simply by creating or collecting data with different classifications.