The recent growth and dependence on Social networks is evident as it helps people opinion on diverse subjects, generating massive data. The analysis of this data is a complex activity. Data mining has many techniques for detecting trends and patterns from huge datasets [1] and Web composition on these complex data sets is an uphill task. Composing services need to leave traditional frontiers and compose for different situations. Web 2.0’s related technologies make it easier to collaborate. A social dimension to these compositions can help satisfy user requirements and thus generate knowledge from user interactions. This paper introduces a Novel Semi Automatic Web Composition for Users Social Interactions (SAWCUSI) from a user’s social perspective.  It discovers and selects based on  user interactions  in a social network.