Space Correction

Space Correction Service

In many languages around the world (Persian, Arabic, etc.), the space character is used to identify words and distinguish them from each other.

Since in Persian, word boundaries are determined by spaces, if spaces are not used correctly, the text becomes unreadable and its meaning is lost or changed.

In Persian, according to the existing standard, there are two types of spacing between words:

• Space: The gap between words in a sentence
• Half-Space (Zero-width non-joiner / ZWNJ): The gap between parts of multi-part words

When spaces and half-spaces are used correctly in texts, several Persian natural language processing tasks—such as spell checking, Persian search engines, Persian text summarization, keyword extraction, machine translation, and indexing—improve significantly.

Example Errors

For example, words like «جمهوری اسلامی» (Islamic Republic) might be incorrectly written as «جمهوریاسلامی», or «نرم افزار» (software) might be written as «نرمافزار».

Solution

The Intelligent Processing Group of the Noor Computer Research Center for Islamic Sciences has launched an intelligent machine-based space correction service. This service can detect correct spaces and half-spaces between words and correct errors. Therefore, if a text contains words like «جمهوریاسلامی», the system will correct it to «جمهوری اسلامی».