User reviews on software platforms offer direct evaluations and attitudes of the software packages, serving as invaluable sources for assessing software quality. However, the large amount of noise, as well as the time-consuming processing of software user reviews, prevents software engineers from effectively utilizing this resource. This work introduces ISOftSentiment, a novel hybrid sentiment analysis tool that addresses this problem and helps software engineers efficiently evaluate what other users think about the quality of software packages from software user reviews. It enhances software quality assessment by integrating sentiment analysis with ISO/IEC 25010:2023 Established Product Quality Model and a fuzzy consensus mechanism. We selected four candidate models for sentiment analysis. These models were trained with 4,062 user reviews from three software review platforms. After an extensive evaluation of the models, we found that the fine-tuned BERT-BiLSTM model achieved the best score with 0.814 accuracy, 0.806 F1, 0.813 precision, and 0.810 recall. Utilizing this model, ISOftSentimen analyzed 51,178 review sentences. The quality analysis outcomes are visualized as a matrix to show the relevant quality attributes of particular software packages. ISOftSentiment provides an approach to assess the quality of specific software packages and visualization of detailed quality attribute analysis through a sentiment matrix, helping software engineers make informed software package selection decisions.