Opinion Retrieval in Twitter
Zhunchen Luo, Miles Osborne, Ting Wang · Proceedings of the International AAAI Conference on Web and Social Media · 2021
We consider the problem of finding opinionated tweets about a given topic. We automatically construct opinionated lexica from sets of tweets matching specific patterns indicative of opinionated messages. When incorporated into a learning-to-rank approach, results show that this automatically opinionated information yields retrieval performance comparable with a manual method. Finally, topic-related specific structured tweet sets can help improve query-dependent opinion retrieval.