Detecting Temporal Sentiment-Oriented Difference for Crowdsourced Time-Sync Comments

Abstract

Recently, with the rapid development of online shared media, a novel time-sync comment occurs. Time-sync comment is a type of crowdsourced user review embedded in online video website and provides better real-time user interaction than traditional mode. User is probably willing to write comments and then send as time-sync comment when reading interested time-sync comments by others or watching attractive video shots. In this paper, we propose temporal sentiment-oriented difference graph (TSoDG) to detect different areas which exist sentiment-oriented difference over time. We attain the TSoDG by calculating the sentiment-oriented score of time-sync comments, and then use semantic similarity to deal with lagged comments. Finally, we analysis the temporal difference in graph by rule. The result can be regarded as a baseline for further research.