Machine-Learned Feature Recommendations From Sponsored Usage Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Online document systems lack access to information about feature usage by non-subscribed entities, limiting the accuracy of feature management and preventing potential customers from being reached.
Innovation Solution
Implementing feature access control and a machine learning model to gather and analyze feature usage data, allowing subscribed entities to sponsor non-subscribed entities and set access policies, while training models to improve feature recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the online document system only tracks feature usage by subscribed entities, then the system maintains simple access control, but the accuracy of feature management and recommendation quality deteriorates
Solution Approach 1:
The patent segments access control into multiple levels: subscribed entities, non-subscribed entities, and sponsored entities. This segmentation allows the system to track feature usage across different entity types independently, improving measurement precision without overwhelming the system with unified complexity. Each segment has its own access rules and tracking mechanisms.
Solution Approach 2:
The patent introduces sponsored entities as intermediaries between subscribed and non-subscribed entities. The subscribed entity (sponsor) grants access to the non-subscribed entity (sponsee) through the sponsorship mechanism. This intermediary structure enables the system to track feature usage by non-subscribed entities while maintaining access control policies defined by subscribed entities, thereby improving feature management accuracy without directly complicating the core access control system.
2Reliability
If the system provides detailed feature access control and sponsorship capabilities, then feature management accuracy improves, but the system complexity increases
Solution Approach 1:
The sponsorship mechanism serves multiple functions simultaneously: it enables feature sharing, tracks usage data, enforces access policies, and facilitates conversion from non-subscribed to subscribed entities. This multi-functionality improves reliability by providing comprehensive feature management capabilities without proportionally increasing system complexity, as a single sponsorship framework handles multiple objectives.
Solution Approach 2:
The system implements feedback loops where feature usage data from sponsored entities is collected and analyzed to improve recommendation accuracy. The machine learning models continuously learn from sponsorship patterns and feature usage behaviors, enhancing system reliability over time. This feedback mechanism allows the system to adapt and improve without requiring fundamental structural changes.
3Measurement precision
If the system collects comprehensive feature usage data from non-subscribed entities, then recommendation quality improves, but data privacy and security requirements increase
Solution Approach 1:
The patent implements local quality by allowing subscribed entities to define specific access policies for different features granted to non-subscribed entities. Each feature sponsorship can have customized permission levels and tracking scopes. This enables precise control over what data is collected from non-subscribed entities, improving measurement precision for relevant features while minimizing unnecessary data collection and associated privacy risks.
Data Source
AI summary
An online document system provides a recommendation for one or more features within the online document system to an entity. The online document system accesses a set of feature training data to train a machine learning model. The set of feature training data may describe characteristics of entities associated with the online document system and historical activity associated with the entities' usage of the online document system's features. The machine learning model may be configured to identify a feature to recommend to an entity based on the entity's characteristics and history of using other features within the online document system. For example, data representing the entity's user accounts and use of an electronic signature feature is used by the machine learning model to identify a document authentication feature to recommend to the entity. The online document system may then provide the identified feature in a recommendation to the entity.