An Effective Hybrid Stochastic Gradient Descent for Classification of Short Text Communication in E- Learning Environments

Nowadays, various media platforms play an inevitable role in regular life in such a way that we couldn't even think of living without these digital platforms. These digital platforms including e-learning systems are excellent for disseminating knowledge such as text, photos, and other pieces of data. Topic classification analysis on these systems have recently attracted a great deal of attention and this is expected to continue. It is the use of non-traditional media platforms that distinguishes them from other forms of communication. As a result, there is an urgent need for efficient methods of assessing the vast quantity of token variants that occur on a regular basis in the digital world. A piecewise Stochastic Gradient Descent (SGD) classification-based algorithm for categorization is proposed in this study. For the representation of textual features, the TF -IDF term weighting system with unigram, bigram, and trigram is used. To boost the effectiveness of the proposed system, partially ordered microword representations of tweets with changing look ahead distances are used. The proposed model is simulated with partial order microwords representation of tweets having lookahead distance 1 and it achieved an enhanced accuracy of 90.73%.