Detection of Topic from Unstructured Text With Mixed Languages

This paper proposes a design of a Topic Detector machine which combines the power of LDA and Word2Vec to detect topic from mixed text. The experiment is carried on a mixed text of English and Hindi to detect topics. The technique tokenizes the mixed text of Hindi and English and models them into feature vector trough a process of Word2Vec. These vectors are clustered and the cluster centers are identified as the topic of the cluster of tokens