Beyond mashups: Graph transformations on web data

The web provides tremendous amounts of data available in various formats, yet processing this data still has many questions unsolved. Various mashup tools were presented in the previous decade to solve this heterogeneity. However, despite the expectations, they have not become very wide-spread. On the other side statistical tools are mostly incapable of directly processing data from the web. In this paper we present an integrated approach to acquire and process web data based on graph transformations. This approach is useful when the data sources are of various formats, and are subjects of basic analysis algorithms like the inventory of qualitative analysis. A case study is presented where Q&A texts are analyzed and user sentiments are estimated to compare similar technologies.