In-context word prediction using on-screen candidate display

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing predictive text input methods for electronic devices require users to divert attention from the keyboard to view and consider suggested words, leading to eye strain, inefficiency, and wasted processing cycles, as they often only provide predictions for the current word being entered and lack contextual understanding.

Innovation Solution

A method that allows users to input text without diverting attention from the keyboard by displaying candidate predictions on the touchscreen, enabling continuous touch input to accept predictions, and using a timer based on touch distance from the screen edge to confirm selections, thus allowing simultaneous selection and confirmation of multiple words.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If predictive text input methods display suggested words on the touchscreen, then text input speed is improved, but users must divert attention from the keyboard causing eye strain and inefficiency

Engineering Contradiction:
Improvetext input speedVSAvoiduser attention requirement
Core Design Contradiction:
ProductivityVSEase of operation

Solution Approach 1:

The patent applies dimensionality change by moving the prediction display from a separate visual location (touchscreen) to the same spatial plane as the keyboard input. Predictions are displayed directly on the keyboard interface, allowing users to select predicted characters without shifting their gaze or attention between different display areas. This resolves the contradiction by maintaining high input speed while eliminating the need for diverted attention.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Loss of energy

If the system provides predictions for only the current word, then processing cycles are optimized, but contextual understanding and prediction accuracy are limited

Engineering Contradiction:
Improveprocessing cyclesVSAvoidprediction accuracy
Core Design Contradiction:
Loss of energyVSMeasurement precision

Solution Approach 1:

The system applies preliminary action by pre-calculating and preparing multiple levels of predictions (current word completions and subsequent word predictions) before the user actually needs them. This allows the system to provide accurate contextual predictions for multiple words ahead of time without wasting processing cycles on unnecessary calculations, as predictions are generated only for relevant contextual positions based on current input state.

Inventive Principle:
Principle #10Preliminary action

3Measurement precision

If users must view and consider suggested words on the touchscreen, then prediction accuracy can be evaluated, but text input efficiency decreases due to attention diversion

Engineering Contradiction:
Improveprediction evaluation accuracyVSAvoidtext input efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent merges the prediction display function with the keyboard input interface by integrating suggested characters directly into the keyboard layout. Users can evaluate and select predictions within the same visual field as their typing activity, eliminating the need to switch attention between keyboard and separate display areas. This combining approach maintains prediction evaluation capability while preserving text input efficiency.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP2639673B1In-context word prediction
Publication Date: 2016.09.14 BLACKBERRY LTD
  • EP2639673B1 patent drawingFigure 1
  • EP2639673B1 patent drawingFigure 2
  • EP2639673B1 patent drawingFigure 3A~3B

AI summary

Methods and systems for predicting user input on a keyboard. Methods include enabling user input on a display comprising at least three fields. The first field displays an input string that is based on input selections such as keyboard entries. The second field displays a candidate prediction generated based on other input selections, consisting at least in part of a proposed completion to the input selection, and partially based on the input string in the first field. The third field displays another candidate prediction generated based on the input string in the first field as well as the candidate prediction in the second field.