Rule based method for entity resolution using distinct tree construction

摘要

Entity resolution is the process of identifying the records that refer to the same entity. Rule based ER works by generating rules from the training dataset obtained from the given dataset and applying these rules to the records in the dataset. This method is very time consuming and tedious as the size of the rule set generated is very large. Also, the rules generated are not efficient enough to classify the records correctly. So a distinct tree construct is proposed to generate the rules from the dataset. Distinct tree is constructed by arranging the dataset in a particular order before rule generation step. Experiments shows that the accuracy of rules generated using distinct tree method is more accurate and fast than simple Rule based ER.