Domain specific text mining almost from scratch with Deep Learning

Deep learning will help us to better design text mining applications, but perhaps not remove the computational linguistic design process associated with text mining applications (Manning, 2015). There has been extensive work on applying deep learning algorithms to different text mining applications such as information retrieval (IR) and information extraction (IE) and so far they have improved on classic IE and IR tasks. However, when deploying the algorithms on more advanced tasks, such as semantic role labelling, there is still some more work to be done (Collobert et al., 2011). In our research we compare and combine traditional natural language processing (NLP) techniques with distributional semantic models for domain specific retrieval.