摘要 |
PROBLEM TO BE SOLVED: To upgrade precision in user action prediction.SOLUTION: An action prediction system 1 that predicts a user's action includes: a transition probability calculation unit 2 that calculates for each movement mode, which signifies each user's use of moving means, a transition probability for a transition from a current visit category, to which each user's current place belongs, to a next visit category, to which the next visiting place belongs, on the basis of the Markov chain dependent on visiting sequence histories of plural users, and a prediction unit 3 that extracts the user's next visiting place from among candidates on the basis of score values of the candidates newly calculated by weighting the scores of the user's next visiting place candidates, which depend on the visiting sequence histories, by the transition probability associated with the movement mode of the user's current place.SELECTED DRAWING: Figure 1 |