A Highly Precise Tap Recognition Method with Behavior-Prediction based on Players’ Input History Analytics
A big problem in the operability of smartphone games is that when a touchscreen is operated with a finger, there is a considerable deviation between the tapped position that is assumed by the player and the position of the tap event recognized by the system. According to this invention, the procedures of operations performed by a group of players are saved as logs, and the game system automatically learns "combinations of operations that are frequently used by the players", thereby realizing a function with which, when a player has performed an ambiguous tap operation, the touch operation that the player intended to perform is predicted from the current operation history of the player to automatically estimate the UI object that is the target of the tap operation. With this invention, in the event of an ambiguous input, such as a case where a finger is placed on multiple objects, it becomes possible for the system to automatically correct the operation or to present the player with a UI that assists reselection of an object. The technical feature of this invention consists in that, in the processing for predicting an operation that a player intended to perform, by using an object in the proximity of a physical tapped position as a search key, operation logs including an enormous number of combinations are quickly narrowed down, which makes it possible to continuously predict and correct tap operations in real time. This invention realizes real-time use and utilization of big data in the field of games, in which high-response performance is required, thereby realizing a basis for a "behavior prediction engine" that predicts the next operation by a player.
Full Text: Google Patents