Abstract
The goal of this article is the development of an evaluation system based on the Arabic language. This article contains four parts. The first part is the corpus construction from the 7th year basic education classes' grammar book in Tunisia. Then, the second part is on the construction of the Concept Maps (CMaps) ontological for simple Arabic sentences from this corpus, where the automatic extraction of terms is completed. This extraction is based on two major approaches: linguistic and statistical. The third part in this article is the automatic instantiation. The last part is devoted to the application of the similarity measure chosen in the CMaps ontological fusion, which summarizes the various semantic links and which ends with a judgment according to the calculated score.