Objective

• To estimate the ratio of Entities in a given text corpus using maximum mean discrepancy.

• Used existing named entity disambiguators for feature extraction and libsvm for benchmarking and comparison.

• Although in the experiments I did, SVM performed better than MMD for ratio estimation of entities, the seminar lead to a useful survey report of the problem of named entity disambiguation techniques and MMD.

Results/Report

• The survey report is available on arxiv.

• Results from the experiments are available here.

• Project presentation