TR-2010006: Graph-Based Clustering and Its Application in Coreference Resolution
Zheng Chen · CUNY Academic Works (City University of New York) · 2010
In this literature review, we survey graph-based clustering and its application in coreference resolution.We state that the methodology of graph-based clustering can be described by a five-part story: (1) hypothesis which hypothesizes that a graph can be partitioned into densely connected subgraphs that are sparsely connected to each other; (2) modeling which deals with the problem of transforming data into a graph; (3) measure which is an objective function that rates the quality of a clustering; (4) algorithm which aims to optimize the measure; (5) evaluation which evaluates the performance of a system clustering relative to a ground-truth clustering.We then survey coreference resolution which is further split into two problems, entity coreference resolution and event coreference resolution.We focus on discussing how the graph-based clustering methodology has been applied in solving these two problems.