Qi Li
University of Illinois, Urbana-Champaign
qili5@illinois.edu
Bio
Qi Li is currently a postdoctoral researcher at the Department of Computer Science at UIUC, working with Professor Jiawei Han. Qi obtained a PhD in computer science and engineering from the University at Buffalo in 2017, advised by Professor Jing Gao, a master’s in statistics from UIUC in 2012. Her research interests lie in the area of data mining with a focus on the collection and aggregation of information from multiple data sources. She has received several awards, including the Presidential Fellowship of the University at Buffalo, the Best CSE Graduate Research Award, and the CSE Best Dissertation from that university’s Department of Computer Science and Engineering.
Truth Discovery from Multi-Sourced Data
Truth Discovery from Multi-Sourced Data
Quality is a major challenge in this big-data era. Data can come from various sources, but those sources are not all equally reliable. It is intuitive to trust reliable sources more but there is often little or no prior knowledge on which piece of information is accurate and which data source is more reliable. For example, in crowdsourcing, human workers may provide conflicting answers, but it is usually unknown who are more reliable and which answers are correct. This is the key challenge in detecting the veracity of information from multi-sourced data. Furthermore, the existence of heterogeneous data types, long-tail distributions, streaming data, unstructured data, etc. further increases the difficulty of discovering the true facts. My research focuses on developing novel unsupervised methods to detect the truths by integrating source reliability estimation and truth finding. Many of my methods have been successfully applied to numerous application domains (for example, health care, transportation, and the environment). They can be incorporated into various research fields (for example, web mining, crowdsourcing, mobile sensing, and information extraction) to improve data quality and achieve better performance.