Representation learning has recently enjoyed enormous success in learning useful representations of semantic data from various areas such as vision, speech, or natural language processing. It has developed at the crossroad of different disciplines and application domains, and as a research field by itself with several successful workshops or sessions at major machine learning and data mining conferences. We take here a broad view of this field and want to attract researchers concerned with statistical learning of representations, including matrix- and tensor-based latent factor models, probabilistic latent models, metric learning, graphical models and also recent techniques such as deep learning, feature learning, compositional models, and issues concerned with non-linear structured prediction models. The first Representation Learning for Semantic Data (ReLSD 2014) Workshop has been held co-located with ECML/PKDD 2014, which received great attentions and participants from many researchers and companies. This year, we seek to hold the Second ReLSD workshop in a more capacious platform as in ICDM 2015. The focus of this workshop will be on representation learning approaches applied to semantic data emerging from real world. New models and learning algorithms based on representation learning that can address all of the challenges in semantic data mining are encouraged. This one-day workshop will include a mixture of invited talks, and contributed presentations, which will cover a broad range of subjects pertinent to the workshop theme. Besides classical paper presentations, the call also includes demonstration for applications on these topics. We believe this workshop will accelerate the process of identifying the power of representation learning operating on semantic data.