MinEmail: SMS Alert System for Managing Critical Emails
Authors: Kyle Rector and Joshua Hailpern
Reviewer: Khawaja Ahmed Ali

1. What is the purpose of the work?

Most of the people rely on email as their primary mode of business communication. However, as number of emails increase, it become difficult to manage them. This may leads to the poor attention to some important emails. Missing an important email may cause delay in project progress and frustration in users. There are numbers of manual and automatic procedures to filter emails based on various features. These methods identify the important emails and highlight them in some way. However, all of the methods have some limitations and thus, an end user cannot get full benefits.
In this paper, authors presented an alert system for emails. MinEmail uses SMS as a notification tool about important emails. It sends an SMS to notify and remind a user about any critical email. This systems automatically predict and label important emails. Authors developed models to predict how and when emails should be addressed. The primary contribution is not to develop a novel system to combine two different communication methods: Email and SMS, but to propose models for prediction and labeling of critical emails.

2. How do the authors achieve this purpose? Why is the particular approach adopted?
Related research shows the three ways of addressing the above mentioned problem: a. Manual Email Management b. Automatic Email Management c. Completion action prediction for email
MinEmail uses completion action prediction models to achieve the purpose. A SMS will be send to user when there is a critical email. Authors defined critical email as an email that is too important to be missed. They use SMS as SMS has quicker time response as compared to emails and is more accessible. The authors presented two scenarios to describe how MinEmail can help an enterprise user to manage his emails. They have also conducted a survey to understand the usage of emails and SMS in an enterprise settings. Based on the previous related work done and results of survey, they created MinEmail, to automatically notify the user about any critical email through SMS. The authors also built server-client architecture. They used MS outlook add-in as their client. They developed a Java-based Apache server for processing and sending emails as MinEmail server. There is only one SMS per critical email and all the alerts are set under a specific threshold. They developed the predictive models based on machine learning techniques and use cross validation method for testing. The authors also conducted an ecological valid study as a follow up survey to determine the impact of MinEmail system in an enterprise settings.

3. Do you think the purpose has been achieved?

The study presented by authors is based on previous work, a 777 person survey and a 3000 emails based experience study. The authors discussed the problem and use of MinEmail by using different scenarios. They presented a quantitative results for initial 777 person survey, which shows that email has higher level of satisfaction and value. They got users among the initial survey for the follow up study. However, due to privacy concerns they cannot track the participants. Due to nature of problem, they conducted study in an uncontrolled environment. The comparative results for four different machine learning predictive classifiers are also presented. Their study shows that 62.19 % of users respond to their important emails in advance, before getting an alert message form MinEmail. Only 121 alerts were sent during study period. Time from 7:00 pm to 8:59 pm was considered as do not disturb time. There is no way to track user behavior after sending an alert. The follow up survey shows that MinEmail is an easy to use and effective system. However, some of the study limitation such as uncontrolled environment, time and privacy limitations are important factors in getting some real time results.

4. What insights have you gained from reading this work?

In this paper, authors describe the two different scenarios to illustrate the problem, if a user miss any critical email and impact of MinEmail for the resolution. The three different ways of managing emails are discussed briefly. The survey shows that users rely on emails as their primary way of communication in an enterprise. Emails have more usability. However, SMS have quicker response time. This paper provides a better understanding of Outlook add on as client and java apache server in a client server environment. This paper also discusses the machine learning techniques as predictive classifiers. The follow up study depicts that different days have different number of critical emails received by an enterprise’s employee and timing also matters in this case. Authors also mentioned that MinEmail cannot solve the problem all alone. Addition factors such as social solution would also be required.