Quality of claim metrics in social sensing systems: A case study on IranDeal

There is an ongoing trend in social sensing where people act as sensors and report the events happening in their surroundings. These claims are often reported by smartphones and need to be processed to discover new patterns of events. Since these claims are not generated with consistent quality, the processing and evaluation tasks can become a challenge. In this paper, we address questions on how the quality of each claim can be evaluated, and which factors should be considered to qualify the quality of the claims. To do this, we investigate the sources of low-quality claims an propose a new form of Quality of Claim (QoC) metrics. We categorize the Quality of Claim factors into two classes of Content Measure and Feedback Measure. The study is performed on Two datasets. The main dataset is the #IranDeal extracted from Twitter. To compare the quality metrics, a second dataset is crawled from the Fouresqure social network. The metrics follow the power law pattern and are modeled by a Zipfian distribution function. The results show the power degree varies from 1.75 to 5. A number of factors are discussed as an influencer of the variation, such as the query criteria of the extracted dataset, the characteristics of the QoC metric, and the type of the social network.