In my research I am currently struggling with the following question. Suppose we represent a text in a high dimensional matrix. Each word is a point or a vector in the matrix. The dimensionalioty of the matrix is determined by the words that co-occur with the word in a sentence. The length of the vector is calculated by sterngth of co-occurance. I would like to divide these matrix to several sub-spaces through hyper-planes. BUT the division of the matrix should be conducted by using "context-words" hints that allow me top divide the "semantic space" into different contexts. Is anyone familar with a way to divide the matrix by using these "hints" as boundary points for the sub-spaces?
