ghost
commented
9 months ago TASKS: This first sprint we will focus on:
- [x] Dummy model that spits out Categories: 1. 'Company Business/Strategy' 2. 'Purely Personal' 3. 'Personal but in a professional context' 4. 'Logistic Arrangements' 5. 'Employment arrangements' 6. 'Document editing/checking/collaboration' 7. 'Empty message (due to missing attachment)' randomly for each email. This will be a stub for others to work on. They can SEND TO and RECIEVE FROM.
- [x] Create separate repo under JARVIS that handles the training of our email sorter. We will begin crafting a 'good' sorter based on the enron emails. Here
- [x] Place first edition of sorter in. depending on power draw of sorter it will either be a remote source or local to the device. If it is remote, we may need to expand out into further PBI.
TIMELINE: Timeline for 3 tasks:
- [x] Set up code to facilitate the 7 categories with a dummy random sort: By 2/10
- [x] Set up tests: By 2/11
- [x] Set up ML repo : By 2/12
- [x] Finish first edition of ML model: By 2/16
- [x] Put in ML sorter, depending on computational cost, either a server connection or locally on the device: By 2/19
Acceptance Criteria: Does it return a category? Does it return results consistent with testing parameters (given some level of accuracy on our test set, I should look for an accuracy of more than 50% on the first model.)
User Story
As Cathie Spino, a research professor and director overwhelmed with managing a heavy load of emails daily, I would like an automated way to sort my emails locally on my device before determining which ones are high priority and need immediate attention. This feature should help me distinguish between critical emails requiring action/response and less important ones, enabling me to focus on urgent matters without the need to manually sift through approximately 250 emails every day. By automatically categorizing emails into 'Informational', 'Requires Action/Response', and 'Less Important', I can quickly attend to those that are most pertinent to my roles as a director and principal investigator, while also efficiently managing communications related to university and potential new projects.
Persona: Cathie Spino, a research professor of biostatistics and director of a unit supporting clinical trials, who is challenged by the task of managing a large volume of mixed-priority emails in her professional and personal life.
Feature: Automated sorting of emails into predefined categories ('Informational', 'Requires Action/Response', and 'Less Important') based on content analysis and user-defined rules, with the capability to learn and adapt to Cathie's preferences over time for more personalized management. [ I WILL ELABORATE ON THIS BELOW ]
Business Value: This feature aims to significantly reduce the time Cathie spends on email management daily, ensuring she does not miss out on critical communications and can make more efficient use of her time. By providing a more effective strategy than current manual sorting methods or the basic sorting features of Outlook, this feature addresses Cathie's need for better control over her email inbox, potentially eliminating the need to hire an expensive secretary. Ultimately, it allows Cathie to focus more on her important work and less on email management.