Matches in ScholarlyData for { ?s ?p Person disambiguation monitors web appearances of a person by disambiguating information belonging to different people sharing the same name. In this paper we extend person disambiguation to incorporate the abstract notion of identity. This extension utilises semantic web technologies to represent the identity of the person to be found and the web resources to be disambiguated as semantic graphs. Our approach extracts a complete semantic social graph from distributed web 2.0 services. Web resources containing possible person references are converted into semantic graphs describing available identity features. We disambiguate these web resources to identify correct identity references by performing random walks through the graph space, measuring the distances between the social graph and web resource graphs, and clustering similar web resources. We present a new distance measure called “Optimum Transitions” and evaluate the accuracy of our approach using the information retrieval measure f-measure.. }
Showing items 1 to 1 of
1
with 100 items per page.
- 198 abstract "Person disambiguation monitors web appearances of a person by disambiguating information belonging to different people sharing the same name. In this paper we extend person disambiguation to incorporate the abstract notion of identity. This extension utilises semantic web technologies to represent the identity of the person to be found and the web resources to be disambiguated as semantic graphs. Our approach extracts a complete semantic social graph from distributed web 2.0 services. Web resources containing possible person references are converted into semantic graphs describing available identity features. We disambiguate these web resources to identify correct identity references by performing random walks through the graph space, measuring the distances between the social graph and web resource graphs, and clustering similar web resources. We present a new distance measure called “Optimum Transitions” and evaluate the accuracy of our approach using the information retrieval measure f-measure.".