evaluate the activity of the students. Apart from this e-learning can be improved if tools for supporting teachers in this task are provided perfectly. Two experiences are presented to show the convenience of providing teachers with such tools5, These experiences proved that providing support to teachers allowed them to assess the students more closely .Result shows that students’ dropout rate in a postgraduate course was reduced and the number of students who passed a physics course in secondary education was increased 5. Drawback of this system is teacher does not usually analyzed activities in e-learning due to lack of data analysis tools 5.The Effectiveness of Personalization in Delivering E-learning Classes6 This system present a methodology of personalized delivery of multimedia resources in an e-learning platform that takes advantage of web 2.0 technologies and its standards, including (i) RSS and Atom feeds are used for streaming(ii) meta-data search engine used for searching (iii) ubiquitous computing is used for delivery purpose(iv) personalized learning System framework has a different phases including:(1) for each student model ,preferred delivery mechanism is considered, (2) measuring the effect of the semantic profile time window parameter, (3)similarity metrics is used for clustering student model (4)result comparison for personalized and non-personalized delivery. This system is the combination of personalization and the data driven extraction of the students models based on similarity metrics 6. Drawbacks this system does not use any method for finding specific domain knowledge and user, also personalization features like availability and user satisfaction was provided6.3. SYSTEM ARCHITECTUREFigure 3.1: System ArchitectureProposed system is developed to improve the recommendation over e-learning platform. In the world of information and communication technology,e-learning has grown to the greater extent with more advantageous way. But at the same time it brings some challenges like, how to learn, how to select a proper teacher over a e-platform. Here, the role of recommender system start. When a student ask a question over forum, teacher responds with the answer to the question. After getting answer from teachers the student gives feedback in the form of ratings to teacher for knowledge. Recommendation system, studies the domain knowledge of the teacher (user) using the previously answered questions and their domains. After obtaining the domain knowledge about of teachers, it finds the availability of the teachers using the number of questions are asked and number of questions answered. At last system obtains the user satisfaction for the particular teacher by using the feedback from the students. Now, recommendation system has the all values which need to be above the respective threshold values. if the values are above threshold values then it recommend that teacher for a student.1.Find the domain knowledge of user:Find the knowledge of user for a domain D by calculating ratio of No. of question answer to No of question asked in that domain.2.Find the user availability:Find the user availability by calculating ratio of Total No. of questions responded to Total No. of questions asked to user i.3.User satisfaction for domain:Find the satisfaction ratio of the user I for domain D by calculating No. of positive feedback to find the satisfaction ratio of the user I for domain D by calculating No. of positive feedback to No. of questions responded by user i.4. Recommended user iif (K(D)I, Ai, Si(p)) ? (KThi, AThi,S(D)Thi)Where,KThi= AThi= S(D)Thi)=0.5IV. SYSTEM ANALYSISIt involves domain knowledge, availability of candidate user and user satisfaction. If these three values are greater than threshold value then user is recommended.Table 1: Data TableV. COCLUSIONThe proposed work implements the user recommendation system using domain knowledge, user availability and the user satisfaction through the reader’s feedback. The system is taking advantage of the all three parameters for finding and recommending a particular user for answering a question.The system can be extended in future to implement the text mining using one of the existing mining algorithms to improve the result of recommendation.ACKNOWLEDGEMENTI am thankful to Prof. N. B. Pokale for his continuous support to complete this work. I am also thankful to Prof. M. K. Kodmelwar his support and guidance during completion of research work. I am thankful to all those who directly or indirectly helped to accomplish this work.