In this paper, we propose a framework for automatic topic modeling to connect the relevant topic of study to the appropriate learners in the e-learning system through by suggesting and recommending the hashtags for those topics of interest. In the field of e-learning the impact of social network is highly influenced due to the facilities like learners interaction, portfolio analysis, content analysis, knowledge sharing, collaborative recommendation, content recommendation and etc. The textual content of the messages, blogs, notes and discussion forums can be extracted in a meaningful way for the purpose of making analysis and to recommend the peer learners participating in the elearning management system. In this paper, we investigated the probabilistic, statistical approach based on the latent dirichilet allocation as the topic modeling method. Through this approach, the set of documents and messages are analyzed for generating and suggesting the hashtags as the representation item for the relevant topics of study. In fact, this proposed approach has been tested on the datasets constructed from the e-learning platform and twitter social media network. The effectiveness of the proposed method has been evaluated using the training and testing approach. © Research India Publications.