Suranah writeup for Borkar 2001

This is a review of Borkar_2001_Automatic_Segmentation_of_Text_Into_Structured_Records by user:Suranah.

The paper attempts to extract structed information from a semi structered, though relatively free form text like citation or mailing address. The authors propose a nested HMM and some modifications to Viterbi algorithm. I especially like the hierarchical feature selection and feel that it can be used for other applications in similar domains which use graphical models. I was a bit disappointed by lack of deeper error analysis.