The term collective entity linking picks up entity linking techniques that take advantage of the interdependency between entities in a body of text to assign simultaneously a set of entities to mentions. In practice, collective techniques can be used alone or in tandem with individual techniques.
Graph is a convenient tool to represent interdependent entities. Very frequently, a graph is defined as nodes representing entity reference and edges representing relations between them. Some researchers include mentions as nodes (e.g. Han et al., 2011) but they can be replaced by scores associated with entity node as well.
Entity reference graph
Entity co-occurrence graph
Page-Rank algorithm (PR)
- Han, X., Sun, L., & Zhao, J. (2011, July). Collective entity linking in web text: a graph-based method. In Proceedings of the 34th international ACM SIGIR conference on Research and development in Information Retrieval (pp. 765-774). ACM.
- Alhelbawy, A., & Gaizauskas, R. Collective Named Entity Disambiguation using Graph Ranking and Clique Partitioning Approaches.