User-generated text reviews (UGTR) have become a crucial source of information for human opinions and business strategies. However, analyzing and extracting valuable insights from such unstructured data presents significant challenges due to its noisy nature and varying quality. In this work, the various pre-processing techniques hold immense importance in the domain of (NLP) tasks when dealing with user-generated reviews. These techniques are crucial to improve the effectiveness and efficacy of our suggested models. The primary objective of this study is to explore different pre-processing techniques by applying two facts (before and after) to make better-quality textual content. This includes tokenization, stopword removal, stemming, lemmatization, case conversion, elimination of special symbols, removal of emojis, hyperlinks, missing words, numeric digits, etc. The primary objective of employing these techniques is to identify patterns, sentiments, and insights within the reviews, which can help businesses better understand customer feedback and refine their products and services for the airline review dataset. Our results demonstrate that careful selection and combination of pre-processing techniques significantly enhance the model performance and reliability of NLP tasks applied to UGTR.