Huang H.-H.Yu C.-H.Chang T.-W.Lin C.-K.HSIN-HSI CHEN2019-07-102019-07-1020149781479941438https://scholars.lib.ntu.edu.tw/handle/123456789/413115Discourse markers not only express some sorts of relations between two arguments, but also entail sentiment information. In this paper, we investigate the associations between the relation type and the sentiment polarity of Chinese discourse markers based on a web scale corpus. We present an approach to mining information from a large scale corpus, show the polarity distributions of sentences under various relation types, and interpret the data from various aspects. Based on the massive amount of data from the Internet, certain language phenomena are shown. ? 2014 IEEE.conference paper10.1109/WI-IAT.2014.842-s2.0-84912568586https://www.scopus.com/inward/record.uri?eid=2-s2.0-84912568586&doi=10.1109%2fWI-IAT.2014.84&partnerID=40&md5=4cc9d7aad557f70a266891602d404169