With the exponential rise in the amount of information in the World Wide Web, there is a need for a much efficient algorithm for Web Search. The traditional keyword matching as well as the standard statistical techniques is insufficient as the Web Pages they recommend are not highly relevant to the query. With the growth in Semantic Web, an algorithm which semantically computes the most relevant Web Pages is required. In this paper, a methodology which computes the semantic heterogeneity between the keywords, content words and query words for web page recommendation is incorporated. A Differential Adaptive PMI Algorithm is formulated for with varied thresholds for recommending the Web Pages based on the input query. The proposed methodology yields an accuracy of 0.87 which is much better than the existing strategies.