Rbalasub writeup of Wang and Cohen - 2008

A review of Wang_2008_iterative_set_expansion_of_named_entities_using_the_web by user:rbalasub

The paper extends the SEAL algorithm to overcome some of the limitations of SEAL. SEAL works well when seeded with 3-4 seeds. Beyond that, the performance drops off. iSEAL counters this problem by using an iterative procedure that repeatedly calls SEAL using automatically extracted entities from the previous iteration. Different rankers including the random walk, bayesian sets and an heuristic wrapper lenght methods are described. The paper also discusses different strategies to choosing the set of entities to be used as seeds in each iteration of the algorithm.