Word-based correction tor retrieval of arabic OCR degraded documents

Arabic documents that are available only in print continue to be ubiquitous and they can be scanned and subsequently OCR'ed to ease their retrieval. This paper explores the effect of word-based OCR correction on the effectiveness of retrieving Arabic OCR documents using different index terms. The OCR correction uses an improved character segment based noisy channel model and is tested on real and synthetic OCR degradation. Results show that the effect of OCR correction depends on the length of the index term used and that indexing using short n-grams is perhaps superior to word-based error correction. The results are potentially applicable to other languages.