Cross-Lingual Sentiment Analysis for Indian Languages using Linked WordNets

Cross-Lingual Sentiment Analysis (CLSA) is the task of predicting the polarity of the opinion expressed in a text in a language Ltest using a classifier trained on the corpus of another language Lt rain. Popular approaches use Machine Translation (MT) to convert the test document in Ltest to Lt rain and use the classifier of Lt rain. However, MT systems do not exist for most pairs of languages and even if they do, their translation accuracy is low. So we present an alternative approach to CLSA using WordNet senses as features for supervised sentiment classification. A document in Ltest is tested for polarity through a classifier trained on sense marked and polarity labeled corpora of Lt rain. The crux of the idea is to use the linked WordNets of two languages to bridge the language gap. We report our results on two widely spoken Indian languages, Hindi (450 million speakers) and Marathi (72 million speakers), which do not have an MT system between them. The sense-based approach gives a CLSA accuracy of 72% and 84% for Hindi and Marathi sentiment classification respectively. This is an improvement of 14%-15% over an approach that uses a bilingual dictionary.