文章基本信息

标题：Truth Finding by Attribute Reliability Estimation for Heterogeneous Data
本地全文：下载
作者：Wenwen Sheng ; Hong Shen
期刊名称：International Journal of Computer and Information Technology
印刷版ISSN：2279-0764
出版年度：2016
卷号：5
期号：2
页码：199-207
出版社：International Journal of Computer and Information Technology
摘要：In the era of big data, data veracity is one of the most challenging problems. One important task in big data integration is to derive the most accurate records from noisy and conflicting data records collected from multiple sources. However, data sources may process a set of properties with inconsistent reliabilities, e.g., height and weight of a patient are more likely to be true than profession in medical records, departure and landing time of a flight are more likely to be true than weather in airline records. In a cloud computing environment, discrepancies among data describing the same object appear more common because of the increased degree of data replication and unknown trustiness of servers storing the data in a cloud. Besides, we observed that the difficulty to provide truth for different entity is quite different. In this paper, we propose an ARTF model to estimate attribute reliabilities with heterogeneous data types and update it with the entity hardness automatically. The property trustworthiness will be more precise in describing source reliability, which in turn will achieve a better precision in inferring the truth. We compare the performance of our method to the state-of-art truth discovery methods through a real world dataset and a synthetic dataset respectively, the experimental results show that our algorithm can process source conflicts much more accurately while reducing the convergence rate.
关键词：truth finding; heterogeneous data types; entity hardness; attribute reliability estimation