Adaptive e-learning as Intelligent Tutoring System have generally been increasing the degree of both intelligence and precision over time as the pressure for personalisation grows and procedures become more standardised.This is gradually pushing any kind of learning contents into the web as a result of widespread of e-learning.And as a result,in literature of the application of e-learning,we can see arguments related to both merits and de-merits.Present study is aimed to explore on improvement the precision of learning preferences and knowledge level classification problems in adaptive learning.An interactive web-based tool was proposed for identifying user differences according to their learning preferences.The accuracy of the tool was presented.A user-generated,microlesson project was developed as an alternative for crowdsourcing.On the project,students created their learning objects in the subject: Introduction to the algorithm.In order to classify user knowledge level,two classifier models are evaluated and integrated to the adaptive system.Finally,their accuracy levels have been compared.For evaluation of the effect size of the proposed adaptive system an experiment was conducted with three instructional groups.Findings suggest satisfactory achievements on control groups.Furthermore,the results lead to suggest positive achievements in the content recommendation based on knowledge level and content difficulty level.Properly addressing of learner preferences lead to enhance student’s behaviour and performance on e-learning systems.