Sgopal1 writeup of Joint inference by poon

From Cohen Courses
Revision as of 11:44, 26 October 2009 by Sgopal1 (talk | contribs) (Created page with 'This is a review of the paper reviewed paper::Poon_2007_joint_inference_in_information_extraction by reviewer::user:sgopal1. All the previous approaches to entity recogn…')
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigationJump to search

This is a review of the paper Poon_2007_joint_inference_in_information_extraction by user:sgopal1.

All the previous approaches to entity recognition, perform segmentation and then identify the duplicate records. This paper proposes a method to do both simultaneously using markov logic networks. They introduce a few rules for capturing the similarity between citations . Combining the two stages in the markov logic network amounts to writing a few more rules. Inference is performed using MC-SAT and weighted voted perceptron. The (overall) results show a 1-2% improvement in performance over other methods. It seems to me like markov logic networks can be used to solve a wide variety of problems. I think I saw a paper that used MLN to compute query sensitive page-rank scores in an online fashion.