We have implemented an Arabic Spell checking API. The API is currently available as a cloud-based web service and it is currently hosted on Amazon Web Services Cloud. This Web API provides suggestions for spelling corrections for selected text taking care of common orthographic and phonological misspelling errors using phonological/orthographic rules. The API uses a huge electronic dictionary of modern standard Arabic containing lexicon of 9 million words driven from modern electronic corpora using Finite State Transducers technique. Finite State Transducers is a technique widely used in the field of Computer Science and Natural Language Processing.   The lexicon of modern standard Arabic contains words currently used by Arab speakers and writers in books, , newspapers, and other media. Our dictionary of Modern standard Arabic includes many words that are not available in the Traditional Standard Arabic. These include words related to politics and science in addition to newly coined terms. The lexicon contains the possible morphological forms (inflections and derivations) for the Modern Standard Arabic words (Example: كتب، ويكتبان، كتبوا، سيكتبن، كاتب، المكتوب).  The lexicon is associated with the feature of suggesting spelling corrections for common orthographic and phonological misspelling errors using context sensitive phonological and orthographic rules. The rules cover a wide range of phonological and orthographic errors. These rules take care of common typos and common errors cussed by the impact if local dialects. For example, our rules cover errors caused by incorrect phonological alternates like changing the correct form تضاعف  to the misspelled form تظاعف  . The rules also deal with errors caused by orthographic typos like changing مسائلto مساءل. Besides, they handle errors caused by the impact of local dialects like and changing خسائر  to خساير.

This Rest API can be used to implement customized spell-checking strategy for a web interface or for a database. It supports two output types: XML and JSON. It can also be used in search components of  web and mobile applications by allowing for the detection of common misspelled queries and the suggestion of correct words. In this case for instance, if the user enters to the web application a misspelled query, the API allows the system to search for the nearest correct spelling word form that has a record in the database of that web application.  Moreover, this API can be utilized in Text Analytics applications to handle Arabic Language.

 

Our Arabic Spell checker system has been recently purchased by Webspellcheker, U.S. company based in Seattle. Our system is integrated now in Webspellchecker spell checking products. This is a demo page featuring our system:

http://information-age-consulting.com/spell