Unsupervised Entity Resolution Method Based on Random Forest

The task of entity resolution is to find records that describe the same entity in the real world, so as to solve the problem of data duplication. This paper proposes an unsupervised entity resolution method based on machine learning. This method first uses LSTM to convert records into vectors with semantic information. Next, we use the improved random forest method to map the records into the n-dimensional space to realize the partition operation of the records, and consider that the records in the same partition point to the same entity. Finally, we use an improvedAffinity Propagation Clustering (AP) to cluster the partitions to determine whether the records in different partitions point to the same entity. Through experiments on real data sets, the effectiveness of the algorithm for solving entity resolution tasks is proved.