What is it about?
Text-pattern matching demonstrates better performance of automated email answering, which is a rule-based text categorization task, compared to matching of document-size bags of words. We show how good the performance can be, and motivate the success of the text-pattern matching by the logical structure of email messages, which is surprisingly uniform. We analyse why text pattern matching may fail and what we can do about it.
Featured Image
Why is it important?
Near term co-occurrence (in one sentence, or in a paragraph) is a surprisingly powerful feature in text matching, and most often an overlooked feature. We demonstrate the power of near term co-occurrence, enhanced with context-dependent synonymy.
Read the Original
This page is a summary of: Automated email answering by text-pattern matching: Performance and error analysis, Expert Systems, October 2017, Wiley,
DOI: 10.1111/exsy.12251.
You can read the full text:
Contributors
Be the first to contribute to this page