This paper develops a new principled framework to solve an emotional-aware truth discovery problem in social sensing applications. Social sensing has emerged as a new application paradigm of cyber-physical systems with humans-in-the-loop where a large crowd of social sensors (humans or devices on their behalf) are recruited to or spontaneously report observations about the physical environment at scale. A fundamental problem in social sensing applications lies in ascertaining the correctness of the reported observations (often called claims) and the reliability of data sources. We refer to this problem as truth discovery. While significant efforts have been made to address the truth discovery problem, an important aspect of the problem has not been fully explored in previous studies: how to deal with emotional claims. A common assumption made in the previous works is that all claims are assumed to be factual (i.e., either true or false). However, unlike physical sensors, humans are more likely to incorporate personal emotions and sentiments in the reported observations (e.g., tweets, blogs), which can easily confuse the current truth discovery solutions and lead to inaccurate results. In this paper, we develop a new emotional-aware truth discovery scheme that explicitly incorporates emotional information of human reported data into an analytical framework. The new truth discovery scheme solves a maximum likelihood estimation problem to determine both the claim correctness and the source reliability. We compare our emotional-aware scheme with the state-of-the-art baselines through three real world case studies using Twitter data feeds. The evaluation results showed that our new scheme outperforms all compared baselines and significantly improves the truth discovery accuracy in social sensing applications.
展开▼
