Literature Review

In order to find datasets (or even ready-to-use learning scenarios) we performed a literature review covering the main conferences and journals. We looked at publications of the respective conference or journal and tried to judge whether the data used in the paper's evaluation section could be useful in the Structured Machine Learning context. Thus, we applied a simple classification scheme, marking datasets (or the respective publication) as not (), maybe () and likely () useful. In a second round we had/will have a closer look at those datasets that are likely or maybe useful and try to construct a meaningful learning scenario. In case we succeed we mark those datasets as done (), or, if it turns out, they cannot be used, with 'no' ().