This research paper presents the conclusive findings of the NEFUSI project. Specifically, it focuses on developing neurofuzzy models capable of accurately and efficiently evaluating semantic textual similarity. Our study reveals that neural networks and fuzzy logic possess distinct characteristics that render them suitable for specific problem domains and unsuitable for others. Neural networks, for instance, excel in pattern recognition yet necessitate user-friendly decision compliance. Conversely, fuzzy logic systems permit interpretation but cannot automatically derive decision-making rules. Recognizing these limitations, we have devised an innovative, intelligent hybrid system combining both approaches to overcome individual constraints simultaneously.