<p>With the tremendous growth in learning approaches, the pre-trained models have acquired immense effect in Natural Language Processing (NLP) development. An improved high-quality summarization has to be concentrated towards the given document, and the similarity among that text (source content) needs to be validated. This research work introduces a novel by fusing the information content of the text and establishing the relevance among the text by tuning the textual content. Here, Fuzzy rules are generated for handling the multiple features of the text content by fine-tuning the text and given as the input to the Convolutional Neural Networks (CNN) to summarize (classify) the text. This model is termed as <i>Fuzzy Tuning for Text summarization with CNN model (FTTS-CNN</i>). The rules are generated to extract the textual content keywords and fuse them into the source content. Next, the semantic similarity metrics are merged with the rule generation module to extract the similarity sources among the given input. Third, the extracted similarity content is provided to the CNN model, where it classifies the text based on higher significance. This tuning process improves the quality of text summarization with high score information for easier access. The simulation process is carried out in MATLAB environment, and the evaluation is performed with various existing approaches. The outcomes generate better trade-offs among the prevailing methods.</p>