Recommender Systems have emerged as an imperative research domain ever since the explosion of information on the web made it impractical to review the exhaustive data in search of specific/valuable content. The application of this technique in various e-commerce related fields have exposed several downsides related to the process through which the online user profiles are evaluated, the semantics of the related content are matched, and the way in which the recommendations rely on the underlying filtering technique. Extensive research in this field has proposed extensions like alleviating the sparsity problem using trust metrics, incorporating a confidence value in the formed similarities among users, studying the privacy/accuracy trade-offs, classifying the database items into attribute class, etc. In an effort to extend the scope of this field, we hereby endeavor to propose a better administration of recommendations through intelligently formed user models. We implement the concept of Most Influential User group [MIU] governed recommendations and hence prove that the quality of decisions that such decision making systems produce better approximates the requirements of the participating users. It also formalizes the new item recommendation model that proposes to alleviate the quality of items modeled in the data warehouse.