Conventional techniques have produced less accurate and lower-quality search results. In order to improve prediction quality, this topic is addressed and a suggested strategy utilizing the supervised learning method is given. A unique paradigm for relevant product suggestions in social networks is presented in this study. This strategy's main objective is to let computers learn on their own, without human help, and then adjust operations as needed. In order to reduce noise, social inputs for suggested method are the first preprocessed. After that, feature extraction is done using an information-based Fisher discriminant approach. The features are then selected from the obtained ones employing the process. Subsequently, product is proposed by employing the supervised learning method to anticipate these clusters. Lastly, exception outputs were assessed, performance measurement looked at, and comparison with traditional approaches were done to confirm the efficacy of the suggested system. The recommended strategy gets a 90.15% positive score out of 3000 evaluations. The analysis shows that the recommended approach is more successful in producing better outcomes for relevant product recommendations.