Review of the main approaches to automated email answering

Abstract

There were 108.7 billion business emails sent daily in 2014, many of them to contact centers. A number of automated email answering techniques have been explored in order to ease the burden of manual handling of the messages. Most techniques stem from three text retrieval approaches – text categorization by machine learning, statistical text similarity calculation, matching of text patterns and templates. The paper discusses the previous research in automated email answering and compares the techniques.