Social media platforms are increasingly becoming a prominent role in the dissemination of information about real-world happenings. So, early identification of newsworthy events from tweets (i.e. feeds from Twitter) is a hot research problem. However, most existing event detection techniques rely primarily on the burstiness of keywords or the changes in structural networks in order to identify events. Often, these techniques fail to see newsworthy events before they reach a trending state due to the tweet’s challenging characteristics and the evolutionary nature of events. Moreover, these methods lack in capturing evolving characteristics of events based on limited contextual information. To address these issues, we propose EventBoost, a window-based tweet processing method for detecting events and associated aspects by exploiting the lexical and semantic affinity among the words in tweets. The event identification method is enhanced by utilising the temporal dimension and social features of tweets, viz. Hashtags and named entities. We use contextual knowledge, in particular, to find semantically similar tweets to form clusters and improve the quality of clusters. The effectiveness of our approach is assessed using standard Tweet Datasets. Our evaluation demonstrates that our approach outperforms the baseline approaches.