Classifying Emails into Human vs Machine Category

TL;DR

This paper introduces a deep learning approach using CNNs to classify emails as human or machine-generated, significantly improving accuracy over previous models and outperforming BERT in this task.

Contribution

The paper presents a novel multi-model CNN framework for email classification, combining content, sender, action, and salutation signals, and demonstrates its superior performance over existing models.

Findings

Full model improves adjusted-recall from 70.5% to 78.8%.

Full model increases precision from 94.7% to 96.0%.

Outperforms state-of-the-art BERT model.

Abstract

It is an essential product requirement of Yahoo Mail to distinguish between personal and machine-generated emails. The old production classifier in Yahoo Mail was based on a simple logistic regression model. That model was trained by aggregating features at the SMTP address level. We propose building deep learning models at the message level. We built and trained four individual CNN models: (1) a content model with subject and content as input; (2) a sender model with sender email address and name as input; (3) an action model by analyzing email recipients' action patterns and correspondingly generating target labels based on senders' opening/deleting behaviors; (4) a salutation model by utilizing senders' "explicit salutation" signal as positive labels. Next, we built a final full model after exploring different combinations of the above four models. Experimental results on editorial data show that our full model improves the adjusted-recall from 70.5% to 78.8% compared to the old production model, while at the same time lifts the precision from 94.7% to 96.0%. Our full model also significantly beats the state-of-the-art Bert model at this task. This full model has been deployed into the current production system (Yahoo Mail 6).