Knowledge Discovery from Multi-Sourced Data by Chen Ye & Hongzhi Wang & Guojun Dai

Author:Chen Ye & Hongzhi Wang & Guojun Dai
Language: eng
Format: epub
ISBN: 9789811918797
Publisher: Springer Nature Singapore

3.3 CTD Algorithm

In this section, we propose the constrained truth discovery algorithm CTD. We start by introducing the whole framework in Sect. 3.3.1. We then detail the technical solution for component Partition and Reduction in Sect. 3.3.2, and propose the overall algorithm CTD in Sect. 3.3.3. We discuss several important issues about implementing CTD in Sect. 3.3.4 and conduct the performance evaluation in Sect. 3.3.5.

Fig. 3.1Framework overview

3.3.1 The Framework Overview

An illustration of the whole framework is shown in Fig. 3.1. Given the multi-source data and a set of DCs, we first propose Partition, an algorithm to partition the entities into disjoint sets based on their relationships. Then, with an initial estimate of source weights, an algorithm Generation is designed to generate linear constraints for each disjoint set for the ease of optimization. Benefiting from the above steps, the truths towards disjoint independent sets are calculated by minimizing the objective function under the linear constraints. To achieve a more accurate result, an iterative procedure is adopted to simultaneously update the source weights and the truths towards disjoint sets.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.