I wish to understand this research challenge on developing a unified data mining theory/framework. I have read a few papers that tackles this problem using several approaches. My initial idea was to just discuss about ways to perform various data mining tasks as a multi step process but i found that these papers are discussing more in terms of the database structure that can support executing various data mining algorithms. My question is how can I do a review of approaches towards developing a unified data mining model without discussing much from the database perspective but more on the process itself and how can I better frame my research question to deal with for example using both clustering and classification techniques to derive insights from data rather than just using a single step approach.

Any ideas on this is much appreciated.

Data mining contains several tasks such as classification, outlier detection, clustering and pattern mining. There are some relationships between these tasks but they are still quite different. Besides, there is a general process for data mining which is to select, prepare and transform the data, apply some data mining algorithms, and visualize/analyze the results.

I think that it is easier to define a unified model for a task in data mining such as pattern mining rather than for all data mining. For example, a unified model for clustering could be a query language that would allow to perform queries to find different types of clusters rather than just one type of cluster.

This is my quick opinion about that.