Ontological Services Using Crowdsourcing
Huairen Lin, Joseph G. Davis, Ying Long Zhou · 2010
This paper develops a service for ontology evolution based on crowdsourcing. The approach is demonstrated using OntoAssist, a specially designed semantic search service that is capable of capturing and disambiguating user’s search intent as well as automatically enabling ontology evolution. Successful and consistent ontology evolution often requires large amount of input data to specify new terms or changes in relationships. These inputs typically come mainly from domain experts or ontology professionals, which makes it hard to keep up with the change of open, dynamic World Wide Web environment. By integrating OntoAssist with an existing search engine, we show that users’ search intent can be disambiguated and aggregated to help to evolve underlying ontology. The disambiguation feature helps the users to find desirable search results. OntoAssist has been implemented and tested by Turkers from Amazon Mechanical Turk in a live demonstration site. Promising results and analysis are reported.