Research work published recently on IEEE Xplore.
A unique aspect of the system is its use of weighted relationships between features—where the weights reflect the importance and strength of inter-feature connections. These weights are incorporated into the final scoring mechanism, allowing the model to more accurately rank solutions based on how well they match the user’s problem context, not just by similarity but by contextual relevance.
The system outputs one or more recommended solutions, each with a correlation score, offering flexibility in decision-making. It was trained on 1,300 curated records and evaluated against 270 real-world use cases. Extensive benchmarking showed that the model outperformed previous solution recommenders, both in terms of semantic relevance and ranking accuracy.