Abstract: Predicting stock prices remains an important subject of big data analytics. Although many prediction models are developed in the literature, the accurate prediction of Stock-Prices is uncertain due to the underlying problem of massive amounts of data with high response time. Hence, for accurate prediction, an ESSM-GRU-based framework is proposed in this paper. Initially, the Twitter dataset is processed to separate automated twits, pre-process the separated twits, and extract features using TF-IDF. Meantime, the attributes from the historical dataset were extracted and merged with the TF-IDF features using the CK-Means-based clustering phase.