Combining Clustering with Canonical Correlation Analysis for Cross-Language Patent Retrieval
Yaoyong Li and John Shawe-Taylor
In: Learning Methods for Text Understanding and Mining, 26 - 29 January 2004, Grenoble, France.
Kernel Canonical Correlation Analysis (KCCA) is a method of correlating linear relationship between two multidimensional variables in feature space. We applied the KCCA to the Japanese-English cross-language information retrieval and classification. The results were encouraging.