Dimensional reduction and semantic vectorization techniques like LSA, pLSA, LDA and Random Indexing do not take advantage of semantic labeled data like Explicit Semantic Analysis (ESA). I am looking for state of art of supervised semantic analysis techniques like ESA.
I have a dataset with hundreds of thousands of concepts. For each concept there are many associated (short) text descriptions. So I want to use some semantic extraction technique to map text description to some semantic feature space. Next I will process new text description using an information retrieval system.
I have just found Sprinkling technique, that add the context columns to document-word matrix to LSA.
I wonder if there are more techniques mixing corpus discovered and a given semantics. Do you know some alternatives to ESA? May be learning to rank helps but I'm looking for semantic extraction to generate input data for learning to rank