The emergence of social media sources has considerably enhanced interaction and collaboration support in the education field. The integration of such sources can equip learners with channels and skills to acquire knowledge through sharing of information, resources, ideas, as well as expressing opinions and comments during the interactions. The generated contents during interaction support learners' understanding and learning concepts and this consequently can guide the delivery of personalized learning experience. However, the utilization of these contents in the current e-learning systems seems inevitably limited. Besides, the interaction is unreasonable. The aim of this research is to highlight and address the issue of limited interaction by opening new space for more participants in addition to the peers and course teacher. Furthermore, analyzing the generated data during group discussion and social collaboration has helped to understand the learning concept and learner's characteristics to provide a personalized learning feature. The research is proposing a framework for personalized elearning with social collaboration support. The framework will be divided into two techniques namely, adaptive learning and collaborative learning through the use of social media sources. The framework is meant to provide value-added services for both students and teachers through an aggregation of information, resources, services and people. This thesis employs different mining and modeling techniques to extract the required information about learner's characteristics and learning concepts. The characteristics targeted in this work are knowledge level, preferences, learning style and social grouping. The knowledge level is measured on the basis of the richness of the shared messages in relation to the learning concept. The learning style is identified based on the preferable learning objects and modeled using Felder-Silverman Learning Style. The identification of learning style has guided the social grouping of the learners based on social interaction. The knowledge level and social grouping direct the delivery of the personalized learning path. The evaluation of the proposed framework targeted Omani learners by considering two educational government institutions in the Sultanate of Oman (Sultan Qaboos University and Ibri College of Technology). The evaluation has been carried out in the form of a case study and interview resulted in an indication of the enrichment of the contents generated during social collaborations. The valuable information included in the extracted content was very useful in understanding the different characteristics of the learners as well as constructing the domain model. This is an indication of a promising further prospective in the utilization of generated data during collaboration activity especially for the delivery of personalization features