Dual-Stage Email Annotation Filtering for Productivity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current email systems do not effectively filter out annotations that have a low probability of user action, often displaying outdated or irrelevant actionable items, which can clutter the user's interface and reduce productivity.
Innovation Solution
Implement a dual-filtering system that applies both server-side and client-side rules to identify and remove annotations with low probability of user action, such as those related to past dates or spam, ensuring only relevant and actionable items are displayed by parsing email text for actionable items like URLs, dates, and times, and applying filtering rules to distinguish and prioritize actionable content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If all actionable items in email text are annotated, then the user can recognize all potential actions, but the interface becomes cluttered with low-probability annotations reducing productivity
Solution Approach 1:
The system changes the parameter of annotation visibility by introducing a probability threshold. Annotations are displayed or hidden based on their calculated likelihood of being actionable, transforming a binary display state into a probabilistic filtering mechanism that adapts to user needs
Solution Approach 2:
Different annotation treatments are applied to different portions of email text based on local characteristics. Each actionable item is evaluated individually and annotated with different visibility levels according to its specific probability of being user-actionable, rather than applying uniform annotation rules
2Reliability
If annotations are displayed for all extracted actionable items, then comprehensive action recognition is achieved, but irrelevant annotations (past dates, spam) clutter the interface
Solution Approach 1:
The system performs preliminary filtering and evaluation of actionable items before displaying annotations. By pre-calculating the probability of user action and applying filtering rules beforehand, the system eliminates low-value annotations before they reach the user interface, preventing clutter proactively
Solution Approach 2:
The system extracts and removes low-probability annotations from the display set. By separating high-probability actionable items from low-probability ones and only displaying the former, the system extracts the valuable information while leaving behind the cluttering elements
3Productivity
If server-side filtering is applied to annotations, then bandwidth and processing load are reduced, but client-side filtering is needed to handle dynamic user context
Solution Approach 1:
The filtering system is segmented into two independent components: server-side filtering that handles static, universal filtering criteria, and client-side filtering that handles dynamic, user-specific context. This segmentation allows each component to specialize in particular filtering tasks without overwhelming complexity
Solution Approach 2:
The system introduces an intermediary annotation probability calculation layer between email reception and display. This intermediary layer evaluates actionable items and assigns probability scores, serving as a mediator that enables both server-side pre-filtering and client-side context-aware filtering to work together effectively
Data Source
AI summary
A system and method for filtering email annotations is disclosed. An email may be parsed for actionable items and email text associated with those items may be annotated. A server-side filter may be applied to the actionable items to remove annotations that have a low-probability for user action. The email and its associated annotations are then stored on the server. An email client may retrieve the email and its associated annotations from the server and perform render-time filtering. This render-time filtering applies real-time checking for annotations which have a low-probability for user action. The email and its associated annotations are then displayed via a user interface.