Presence Data Affect Estimation Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current presence systems struggle to accurately estimate users' emotional states and communication preferences in technology-mediated communication settings, as existing methods are invasive, costly, or face privacy issues, limiting their adoption and effectiveness.
Innovation Solution
A method that utilizes presence data to derive features from a user's recent presence states, desktop activity, and workflow, applying an estimation model to predict affect and communication preferences without requiring user action or wearable sensors, using a decision tree classifier and under-sampling techniques to improve prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If wearable sensor devices are used to collect physiological information for affect prediction, then measurement precision is improved, but device complexity and cost increase significantly
Solution Approach 1:
The patent extracts only the necessary presence state information from existing presence system data without adding wearable sensors. It takes out the affect prediction function from the complex physiological monitoring domain and implements it using simple presence state transitions already captured by standard presence systems.
Solution Approach 2:
The patent uses inexpensive, non-invasive presence state data instead of expensive wearable sensors. The approach replaces costly continuous physiological monitoring with cheaper, discrete presence state transitions that are already being tracked by presence systems.
2Ease of operation
If camera-based facial expression recognition is used, then affect information can be obtained without user action, but privacy issues arise and adoption is limited
Solution Approach 1:
The patent extracts affect prediction capability from invasive camera-based facial analysis and relocates it to non-invasive presence state transition analysis. This maintains the benefit of automatic affect detection while eliminating privacy concerns by using only voluntary presence system data.
3Loss of information
If text analysis methods are used to infer affect, then affect information can be derived, but significant privacy issues negatively impact adoption
Solution Approach 1:
The patent extracts affect prediction from text analysis that requires accessing private communication content and relocates it to presence state transition analysis. This maintains affect information availability while eliminating privacy violations by using only public presence system data that users voluntarily share.
4Ease of operation
If presence state transition data is used for affect prediction, then user adoption is improved and privacy is protected, but prediction accuracy must be maintained without wearable sensors
Solution Approach 1:
The patent changes the parameters used for affect prediction from physiological measurements to presence state transition characteristics. By analyzing patterns in how users transition between presence states (duration, frequency, sequence), the system achieves accurate affect prediction using only voluntary presence system data.
Solution Approach 2:
The patent implements feedback loops where presence state transitions are continuously monitored and used to update affect predictions. The system learns from patterns in presence state changes over time, improving prediction accuracy through continuous feedback from user behavior without requiring additional sensors.
Data Source
AI summary
Exemplary embodiments described herein are directed to systems and methods that estimate a user's affect and communication preferences from presence data. The exemplary embodiments use a small set of features derived from a user's recent high level presence states. Exemplary embodiments also use features from broader classes of presence data. Utilizing features from a combination of presence data and recent presence states may provide improvement over estimates that users are able to make themselves. The exemplary embodiments further consider cost, burden on the user, and privacy issues in estimating affect and communication preferences.