Web mashup or mashup development is a Web 2.0 software development approach in which users are expected to create applications by combining data sources, application logic and UI components together to cater for their situational application needs. However, in reality creating even a simple mashup application is a complex task and that can only be managed by skilled developers.

Examples of ready mashup models are one of the main sources of help for users who don't know how to design a mashup, provided that suitable examples can be found (examples that have an analogy with the modeling situation faced by the user). Tutorials, expert colleagues or friends, etc. are few other typical means to find development help. However, searching for such kind of help does not always lead to a success, and retrieved information is only seldom immediately usable as it is, since the retrieved pieces of information are not contextual. Motivated by the development challenges faced by a naive user of existing mashup tools, in this research the main focus is to aid less- skilled users in mashup development by enabling assisted reuse of pattern-based composition knowledge. In this talk I will demonstrate how such development assistance can be provided with the help of contextual, interactive recommendations of composition patterns that are harvested from the existing composition models in mashup tools. I will explain in details how the related challenges for this research thread are tackled from the algorithmic as well as from the application perspectives. I will briefly explain the primary assumptions behind this research as well as I will report the result of my research and demonstrate the efficiency of new set of algorithms for contextual pattern retrieval and reuse.