Meta-node Composition Method on Multilingual Parallel-text Graph Having Many-to-many Relationship

Recently, there is increasingly globalized world. We developed a multilingual parallel-text sharing system for multilingual communication. A parallel-text requires one-on-one combination among each language. However, parallel texts often have example sentences of many-to-many combination among each language. A multiplicity of expression causes this problem. It is difficult to solve this problem only information of association between example sentences. Therefore, we proposed a meta-node composition method for parallel-text graph having many-to-many combination. Moreover, we proposed a new parallel-text link discovery method to solve latent problems.