Weibo topics are characterized by diversity and complexity, and often contain different sub-topics even for the same topic. Therefore, how to classify sub-topics effectively and accurately is of great significance. Due to the strong similarity between sub-topics belonging to the same topic, most existing methods cannot be directly applied to the task of sub-topic discovery. In this paper, a new technique based on sentence embeddings for detecting sub-topics of weibo is proposed. The information of blogs is represented by sentence embeddings containing semantic information and the results are verified by clustering.
展开▼
