In this report our main target is multi-source entity resolution in the setting of pair-wise data integration. Unlike the majority of present solutions to entity resolution this study will not consider matching to be transitive. A result of this is that entity resolution on multiple sources will not be confirmed to be associative. The intention of this study ended up being to create a universal model for multi-source entity resolution in the setting of pair-wise data integration that is associative. The primary contributions of this investigation are: (1) an elegant model for multi-source entity resolution and (2) techniques that could be utilized to fix matching conficts in such a way that makes multi-source entity resolution to be associative. The potential worlds semantics is utilized to deal with uncertainty caused by possible matches.
Contents: Multi-source entity resolution
1 Introduction
1.1 Motivation
1.2 Research Question
1.3 Approach
1.4 Overview
2 Related Research
3 Informal problem de nition
3.1 Integrating 2 sources
3.1.1 Matching items
3.1.2 Matching con
icts
3.1.3 Possible worlds semantics
3.1.4 Merging items
3.2 Integrating multiple sources
3.2.1 Matching merged items
4 Formalization
4.1 Matching two items
4.2 Merging two items
4.3 Matching merged items
4.4 Matching two sets of items
4.5 Annotating edges
4.6 Repairing con
icts
4.7 Enumerate possibilities
4.8 Merge items in a matching graph
4.9 Integrate two sets of items Strategies
5.1 Problematic matching cases
5.2 Approach
5.2.1 Cluster items
5.2.2 Dependencies between edges
5.3 All distinct strategy
5.4 All the same strategy
5.5 All possible strategy
5.6 Reconsider strategy
5.7 Comparison between strategies
6 Conclusions and future work
6.1 Future work
A Formalization of an example
Source: University of Twente
Download URL 2: Visit Now