Exploiting contextual information to improve call prediction.

Area of Science:

• Computer Science
• Human-Computer Interaction
• Data Science

Background:

• Mobile phone contact lists are growing, making contact management and retrieval difficult.
• Predicting user call behavior is crucial for efficient mobile phone interaction.

Purpose of the Study:

• To analyze key dimensions for predicting user call behavior.
• To improve existing contact prediction algorithms by incorporating temporal information.

Main Methods:

• Analysis of dimensions contributing to call prediction.
• Enhancement of a state-of-the-art algorithm using frequency, recency, and temporal data.

Main Results:

• The improved algorithm demonstrates better overall prediction accuracy.
• Significant improvement observed in predicting a user's top contacts compared to the base algorithm.

Conclusions:

• Incorporating temporal information enhances predictive models for user calls.
• The refined algorithm offers a more effective solution for contact management and prediction in large contact lists.