Combining N-Grams and Stemming for Arabic Word-Based Inexact Matching and Term Conflation

In this paper, the results of three N-gram techniques have been reported. Two of these techniques were based on the idea of combining N-grams and stemming. The first used first-order stemming, while the other used light stemming. The performance of the combined approach was then compared with that of pure conventional N-gram-based string matching. The results provide good evidence that combining N-grams with stemming improves the overall performance, as measured by word-match recall and word-match precision, using different similarity threshold values.