Sentiment analysis is one of the fastest-growing fields in computer science, favoring machine learning models. Positive, negative, and neutral evaluations may broadly categorize user opinions. Sentiment analysis is the process of analyzing variations in views and categorizing them. Data mining is a technique for extracting meaningful information from online pages, mainly e-commerce networking sites. Combining data mining with other domains, such as text mining, NLP, and artificial intelligence, can categorize reviews as excellent, terrible, or neutral. This research mainly emphasizes the classification of text news review data gathered from the Kaggle Web site. We proposed the stacked ensemble machine learning model for an efficient mining approach for text news assessment. The datasets are taken from the Data World Web site. The datasets are preprocessed using the Lancaster stemming algorithm using NLTK. We improved the training model with the Bi-LSTM with CNN algorithm for better classification results with improved Adam optimization. The best features are imported using RF with LR to select the features and fit a given dataset. Finally, we are using stacking ensemble machine learning (ML) methods to improve the efficiency and reliability of the suggested strategy. The experimental findings show that our suggested strategy outperforms individual classifiers regarding f-measure and accuracy.