cost-sensitive data mining with boosting, bagging or stacking. you may use heart or statlog or credit-card data-set from UCI. Many researchers including me are working on the same. The area is interesting. If you want to find more information on the same. Write to me.
With increasing interest in the Semantic Web and the Resource Description Framework, there is a real need for Data Mining tools and techniques that can extract data triples from a wide range of sources (including traditional databases, electronic documents, web pages and Social Media).
Please, see http://amf21.ru/biblioteka/meroprijatija-provodimye-associaciei/proekt-rffi-sozdanie-bazy-znanii-nanom - nanotechnology area. Also: http://www.chuvsu.ru/2008/proekt.html - education system , http://mfi.chuvsu.ru/opros/ - family relation and marriage duration
There is a wide range of researches in the data mining, which you can use the UCI data sets for them. Technically, you can apply classification, clustering and association rule mining methods, etc. on these data sets. It can be contained from simple decision tree (in classification) or k-means (in clustering) to advance ensemble approaches. I propose studying some main references in data mining such as TAN et al. or HAN el al. Books for finding some simple implementations for your research.