Recognition and Retrieval of Mathematical Expressions from Arabic Documents

Abstract

Recently, researchers have strived to enhance existing search tools to enable the retrieval of a diversity of data types to help users in finding what they are looking for from the Web, however, very limited research is applicable to Arabic content. In this research, we introduce a rule-based approach to search for mathematical expressions written in Arabic and/or English from Arabic stored documents. A set of normalization as well as math expressions' equivalence rules were built to enhance the capabilities of a math search engine. Furthermore, rules for structural search to enable search for sub-expressions were also built. An indexing mechanism and a mapping between Arabic and English expressions during the search process was also produced. The approach was applied using a set of forty queries; written in Arabic and/or English, was applied on a manually collected dataset of 100 documents which has produced an overall accuracy of 75%.