Towards ubiquitous computing with call prediction

Abstract

With the long-awaited era of the pervasive computing approaches, the handheld devices such as personal mobile phones begin to evolve into ubiquitous computing devices. At this early stage of the evolution, we propose a model of a call predictor based on the naïve Bayesian classifier. As an incoming call predictor, our model makes use of the user's call history to generate a list of numbers/contacts that are the most likely to be the callers within the next hour. On the other hand, when the user wants to make an outgoing call (e.g., user flips open the phone or unlocks the phone, etc.), the outgoing call predictor generates a list of number/contacts to be called. Our model has been evaluated with the real-life call logs and it shows a promising result in accuracy.