Soft Keyboard Probability Distribution for Input Accuracy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Soft keyboards on touch-sensitive displays often suffer from poor performance due to the lack of tactile feedback, leading to mismatches between the user's intended key target and the detected key, causing user frustration and erroneous responses.

Innovation Solution

The computing device determines probability distribution functions for each key on the soft keyboard based on user interaction characteristics and keyboard layout parameters, such as typing speed and key spacing, to accurately predict the intended key target, with these probabilities being used to improve input processing and prediction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If a soft keyboard is displayed on a touch-sensitive display without tactile feedback, then the device can be controlled to display a keyboard for input, but mismatches occur between the user's intended key target and the detected key

Engineering Contradiction:
Improvekeyboard input capabilityVSAvoidkey target detection accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The system implements probabilistic feedback by calculating and utilizing probability distribution functions that model the relationship between touch locations and intended keys. The text prediction engine receives these probabilities and provides contextual feedback to resolve ambiguities, effectively creating a closed-loop system that compensates for the lack of tactile feedback.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent introduces probability distribution functions as an intermediary layer between the raw touch detection and key identification. This intermediary computes the likelihood that a given touch location corresponds to each possible key, allowing the system to make informed decisions about user intent rather than relying on simple proximity matching.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If probability distribution functions are computed for each key based on user interaction characteristics and keyboard layout, then input accuracy is improved, but computational complexity increases

Engineering Contradiction:
Improvekey target detection accuracyVSAvoidprobability computation complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system dynamically adjusts parameters of the probability distribution functions based on observed user interaction characteristics such as typing speed and touch location patterns. By changing these parameters adaptively rather than using fixed values, the system achieves high accuracy without requiring excessively complex models for each individual case.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent employs universal probability distribution function models that can be applied to any key on the keyboard through parameter instantiation. Rather than creating specialized complex models for each key, the same functional form is used across all keys with parameters adjusted based on keyboard layout and user behavior, reducing overall computational complexity.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP2756369B1Soft keyboard interface
Publication Date: 2020.03.11 MICROSOFT TECHNOLOGY LICENSING LLC
  • EP2756369B1 patent drawingFigure 1
  • EP2756369B1 patent drawingFigure 2
  • EP2756369B1 patent drawingFigure 3

AI summary

User experience of the performance of a computing device is improved through an operating system that processes inputs from a soft keyboard to provide information that can be used to accurately determine keys a user intended to strike while typing. For each detected tap, the operating system provides a probability that one or more keys were the intended target for the user. These probabilities may be computed from probability distribution functions that are dynamically determined based on user and/or system factors, such as typing rate and keyboard style or layout. Other components may use the probabilities to select a key corresponding to a detected keyboard tap as representing the intended user input. The selection may be made based on the probabilities alone or in combination with contextual factors that yield an overall probability that a detected tap arose from a user targeting a specific key.