RemindMe: Plugging a Reminder Manager into Email for Enhancing Workplace Responsiveness

Casey Dugan, Aabhas Sharma, Michael J. Muller, Di Lu, Michael Brenndoerfer, Werner GeyerINTERACT 2017 · DOI

TL;DR

This paper introduces RemindMe, a novel system that proactively reminds users about pending email requests, reducing cognitive load and improving workplace responsiveness. By integrating a reminder manager directly into email clients via a browser plugin, RemindMe aims to strike a balance between providing timely awareness and avoiding disruptive interruptions.

What Problem Does This Paper Address?

In today's workplace, email has become a primary channel for task management and social requests. This often leads to "email overload," where users struggle to keep track of incoming and outgoing requests, leading to missed deadlines and increased stress. A common strategy to manage this is sending reminders, but this itself can be cognitively demanding and prone to errors. The core problem RemindMe addresses is how to proactively help users manage these email-based requests without adding to their existing cognitive burden or causing disruptive interruptions. The paper highlights the challenge of identifying which emails truly require a reminder and determining the optimal time to deliver it, all while respecting the user's workflow.

Methods

RemindMe employs a multi-component architecture consisting of a browser plugin and a backend server.

  • Email Synchronization: The browser plugin continuously queries the user's email client for both incoming and sent emails, sending this data to the RemindMe server.
  • Data Cleaning and Storage: Incoming emails are processed to strip HTML, signatures, and reply histories, preserving a clean, line-based text representation.
  • Request Identification: A key innovation is the use of a third-party "action identification" API (Watson Workservices) to analyze email text (subject and body) and identify potential requests, questions, or actions. This goes beyond simple spam/ham classification.
  • Conversation Analytics: The system stores representations of conversation threads to understand email relationships.
  • Reminder Creation: This module combines request identification with user-specific reply statistics. It automatically determines if a user is likely to need a reminder about a specific email based on the presence of a request and the user's historical reply patterns to the sender. It also supports manually flagged reminders.
  • Reminder Delivery: The browser plugin injects a dedicated "reminder area" into the email interface. This area presents reminders unobtrusively, allowing users to control when they engage with them. It supports both automatic and manual scrolling between active reminders.
  • Feedback and Learning Mechanisms: RemindMe incorporates both implicit and explicit feedback. Users can mark reminders as "Done" (request fulfilled) or "Remove" (misidentified reminder), or "snooze" them. This feedback is used to tune the machine learning algorithms for better request identification and timing.

Datasets

Not explicitly discussed in the paper. The system is designed to work with existing enterprise email systems and relies on user interaction data and an external API for request identification.

Key Insights and Ideas

  • Proactive Reminders as a Solution to Email Overload: The paper argues that proactive, intelligent reminders, rather than reactive ones, are crucial for managing the cognitive load of email.
  • Balancing Awareness and Interruption: A central theme is the delicate balance between making users aware of pending tasks and avoiding disruptive notifications. The dedicated, non-intrusive reminder area is a key design choice to achieve this.
  • Leveraging User Behavior for Reminder Timing: Using a user's historical reply patterns to specific senders is a clever way to predict when a reminder would be most effective, rather than relying on fixed time intervals.
  • The Importance of Contextual Reminders: Reminders are most effective when they include context about the "trigger event" and the "intended activity," drawing on cognitive psychology research.
  • Feedback is Crucial for Learning: The system's ability to learn and adapt relies heavily on user feedback, both explicit (e.g., "Done," "Remove") and implicit (e.g., snoozing).

Key Results

The paper presents the architecture and design of the RemindMe system. While specific quantitative results from a large-scale deployment are not detailed, the paper describes the system's ability to:

  • Automatically identify emails likely containing requests.
  • Predict the likelihood of a user needing a reminder based on request identification and reply history.
  • Deliver reminders unobtrusively within the email client.
  • Gather user feedback to improve reminder accuracy and timing.

The system is delivered as a browser plugin, allowing integration with existing web-based email systems.

Limitations and Future Work

The authors acknowledge several limitations and suggest future directions:

  • Simplistic Fulfillment Model: The current system assumes a reply to an email signifies request fulfillment, which may not always be true (e.g., "I'll send the file later"). Future work includes incorporating more detailed methods for determining actual request satisfaction.
  • Context-Awareness: The system could be improved by sensing the user's current task or task boundaries to optimize reminder timing.
  • User Feedback Burden: While feedback is crucial, the authors plan to investigate the burden of manual feedback and users' privacy concerns regarding email analysis.
  • Personalized Time Intervals: Further research is needed to develop more sophisticated algorithms for personalized reminder timing based on user feedback.
  • User Study: A detailed user study is planned to evaluate the personalized learning mechanisms and uncover additional user needs and preferences.
  • Broader Integration: While designed for extensibility, the initial focus was on Firefox. Future work could expand support for other browsers and email clients.

Summary updated

BibTeX
@inproceedings{dugan2017remindme,
  author = {Casey Dugan and Aabhas Sharma and Michael J. Muller and Di Lu and Michael Brenndoerfer and Werner Geyer},
  title = {{RemindMe: Plugging a Reminder Manager into Email for Enhancing Workplace Responsiveness}},
  booktitle = {INTERACT},
  year = {2017},
  doi = {10.1007/978-3-319-67684-5\_24},
}
Figure