Automating the generation of questions to assess students remains a serious scientific problem in developing country. Several works have been carried out on the subject but, to date, the automation of the correction of the questions generated, their connection to identified educational objectives, their ability to address comprehension problems, the definition of their difficulty and the response time remain points that are not very much addressed. In developing countries, this problem is even more accentuated by the massification due to overstaffing. This article presents a mathematical logic question generator that provides the generation and correction of questions linked to objectives, testing comprehension, with the possibility of calculating a level of difficulty and automatic scoring using the connectors and quantifiers making up the question. This generator uses automatic language processing techniques based on rules, as well as a formalization of learning objectives allowing correspondence with the generated questions. The test of the generator among mathematics students shows that 100% of the automatically generated questions comply with the objectives for which they were generated. About 90% of generated questions are syntactically and semantically well-formed