A Context-Aware Recommendation Method with Players Behavior Analysis for Game-Screen Videos Captured by Users
This invention provides a context-aware video recommendation method for recommending a play-video having high relevance in accordance with a player's action and status in a game. According to this invention, when a player creates a play-video in a game, an action and status of the player being captured are stored in a server as metadata. The technical novelty of this invention consists in that a situation in which a manner of play by another player is to be presented is detected as an event in a game, and in response to that event, the degrees of relevance between the situation faced by a player and already uploaded play-videos are calculated as distances in a high-dimensional vector space and displayed in the form of ranking. This makes it possible to recommend a play-video that is most suitable for the current status of the player without depending on conventional sharing methods based on PVs or via SNSs. By recommending play-videos seamlessly in accordance with the status in a game, players who create videos will have more chances for their videos to be viewed by many users, and game players can learn new ways of playing the game. With this video recommendation mechanism, an environment for viewing play-videos of a game is implemented within that game. Therefore, by applying this invention, it becomes possible for a company that provides game software to fully control ranking and listing of videos.
Full Text: Google Patents