The proliferation of data on the internet has been greatly accelerated by the emergence of social media platforms over the past twenty years. These platforms serve as valuable sources of user-generated information, with Twitter particularly standing out as a popular microblogging platform that provides concise insights. However, analysing informal expressions from such platforms presents significant challenges, especially in understanding and analysing customer concerns within the telecom sector. Our research focuses on pre-processing natural language sentences to aid comprehension and analysis. We explored two distinct approaches: using the Universal Sentence Encoder and pre-processing models to prepare tweets for analysis. Additionally, we utilized algorithms such as BERT and regression after pre-processing. This approach allowed us to test four distinct modules: pre- processing with BERT, pre-processing with regression, Universal Sentence Encoder with BERT, and Universal Sentence Encoder with regression. The categorized data is then leveraged to develop predictive models through machine learning techniques aimed at assessing public sentiment and anticipating customer issues within the telecommunications sector. This study seeks to advance pre-processing methodologies to improve decision-making processes and enhance customer satisfaction within the telecommunication industry segment.