While the API is great for text mining for the provided taxonomy in the news space, I have a need where I would like to text-mine from a corpus in a particular domain; this calls for me to define my own taxonomy, and extract information (entity, relationships) in this particular domain/taxonomy.

Do you provide the core NLP engine, + the basic entity/relationship extraction, and in addition provide a language/script using which I can build my own semantic layer and extraction of information in my domain of interest?