Method and system for prophesying behavior of online gamer, and computer program product thereof

Abstract

A method and a system for prophesying a behavior of an online gamer, and a computer program product thereof are provided. In the present method, a gamer descriptor of a specific gamer of an online game accumulated before a time point is obtained firstly. Then, the gamer descriptor is divided into a plurality of sub-traces according to the time. Thereafter, at least one feature is derived from each of the sub-traces, and all features derived from the sub-traces are used for determining a possibility that the specific gamer quits the online game within a specific time period in the future.

1 . A method for prophesying behavior of an online gamer, suitable for prophesying a behavior of a specific gamer of an online game at a time point, comprising:
 obtaining a gamer descriptor of the specific gamer accumulated before the time point;   dividing the gamer descriptor into a plurality of sub-traces according to time;   respectively deriving at least one feature value from each of the sub-traces; and   determining a possibility that the specific gamer quits the online game within a specific time period in the future according to the at least one feature value derived from each of the sub-traces.   
     
     
         2 . The method for prophesying the behavior of the online gamer as claimed in  claim 1 , wherein the gamer descriptor at least comprises one of a game related record of the specific gamer and a life related record of the specific gamer. 
     
     
         3 . The method for prophesying the behavior of the online gamer as claimed in  claim 2 , wherein the game related record at least comprises one of an online/offline record, a character profile information, and a character behavior information. 
     
     
         4 . The method for prophesying the behavior of the online gamer as claimed in  claim 3 , wherein when the gamer descriptor comprises the online/offline record, the step of respectively deriving the at least one feature value from each of the sub-traces comprises:
 respectively deducing at least one of an average daily playtime, a playing density, an average session time point, an average of each session time, and a variation of daily playtime corresponding to each of the sub-traces according to each of the sub-traces, so as to serve as the at least one feature value.   
     
     
         5 . The method for prophesying the behavior of the online gamer as claimed in  claim 2 , wherein the life related record at least comprises one of a gamer profile information and a gamer behavior information. 
     
     
         6 . The method for prophesying the behavior of the online gamer as claimed in  claim 1 , wherein the step of determining the possibility that the specific gamer quits the online game within the specific time period in the future according to the at least one feature value derived from each of the sub-traces comprises:
 using a machine learning mechanism to process the at least one feature value derived from each of the sub-traces, so as to calculate the possibility that the specific gamer quits the online game within the specific time period in the future.   
     
     
         7 . The method for prophesying the behavior of the online gamer as claimed in  claim 6 , wherein the machine learning mechanism comprises one of a supervised learning classification and a non-supervised learning classification. 
     
     
         8 . The method for prophesying the behavior of the online gamer as claimed in  claim 7 , wherein the supervised learning classification comprises a support vector machine (SVM). 
     
     
         9 . A system for prophesying a behavior of an online gamer, comprising:
 an input/output interface;   a storage unit, configured to store gamer descriptors of a plurality of gamers of an online game;   a feature deriving unit, coupled to the input/output interface and the storage unit, wherein after the input/output interface obtains a specific gamer and a time point, the feature deriving unit is configured to obtain the gamer descriptor of the specific gamer that is accumulated before the time point from the storage unit, divide the gamer descriptor into a plurality of sub-traces according to time, and respectively derive at least one feature value from each of the sub-traces; and   a prophesying unit, coupled to the input/output interface and the feature deriving unit, the prophesying unit is configured to determine a possibility that the specific gamer quits the online game within a specific time period in the future according to the at least one feature value derived from each of the sub-traces, and output the possibility through the input/output interface.   
     
     
         10 . The system for prophesying the behavior of the online gamer as claimed in  claim 9 , wherein the gamer descriptor at least comprises one of a game related record of the specific gamer and a life related record of the specific gamer. 
     
     
         11 . The system for prophesying the behavior of the online gamer as claimed in  claim 10 , wherein the game related record at least comprises one of an online/offline record, a character profile information, and a character behavior information. 
     
     
         12 . The system for prophesying the behavior of the online gamer as claimed in  claim 11 , wherein when the gamer descriptor comprises the online/offline record, the feature deriving unit respectively deduces at least one of an average daily playtime, a playing density, an average session time point, an average of each session time, and a variation of daily playtime corresponding to each of the sub-traces according to each of the sub-traces, so as to serve as the at least one feature value. 
     
     
         13 . The system for prophesying the behavior of the online gamer as claimed in  claim 10 , wherein the life related record at least comprises one of a gamer profile information and a gamer behavior information. 
     
     
         14 . The system for prophesying the behavior of the online gamer as claimed in  claim 9 , wherein the prophesying unit uses a machine learning mechanism to process the at least one feature value derived from each of the sub-traces, so as to calculate the possibility that the specific gamer quits the online game within the specific time period in the future. 
     
     
         15 . The system for prophesying the behavior of the online gamer as claimed in  claim 14 , wherein the machine learning mechanism comprises one of a supervised learning classification and a non-supervised learning classification. 
     
     
         16 . The system for prophesying the behavior of the online gamer as claimed in  claim 15 , wherein the supervised learning classification comprises a support vector machine (SVM). 
     
     
         17 . A computer program product comprising a plurality of program instructions, the program instructions being loaded to a computer system to execute following steps:
 obtaining a gamer descriptor of a specific gamer of an online game that is accumulated before a time point;   dividing the gamer descriptor into a plurality of sub-traces according to time;   respectively deriving at least one feature value from each of the sub-traces; and   determining a possibility that the specific gamer quits the online game within a specific time period in the future according to the at least one feature value derived from each of the sub-traces.   
     
     
         18 . The computer program product as claimed in  claim 17 , wherein the gamer descriptor at least comprises one of a game related record of the specific gamer and a life related record of the specific gamer. 
     
     
         19 . The computer program product as claimed in  claim 18 , wherein the game related record at least comprises one of an online/offline record, a character profile information, and a character behavior information. 
     
     
         20 . The computer program product as claimed in  claim 19 , wherein the program instructions respectively deduce at least one of an average daily playtime, a playing density, an average session time point, an average of each session time, and a variation of daily playtime corresponding to each of the sub-traces according to each of the sub-traces, so as to serve as the at least one feature value if the gamer descriptor comprises the online/offline record. 
     
     
         21 . The computer program product as claimed in  claim 18 , wherein the life related record at least comprises one of a gamer profile information and a gamer behavior information. 
     
     
         22 . The computer program product as claimed in  claim 17 , wherein the program instructions use a machine learning mechanism to process the at least one feature value derived from each of the sub-traces, so as to calculate the possibility that the specific gamer quits the online game within the specific time period in the future. 
     
     
         23 . The computer program product as claimed in  claim 22 , wherein the machine learning mechanism comprises one of a supervised learning classification and a non-supervised learning classification. 
     
     
         24 . The computer program product as claimed in  claim 23 , wherein the supervised learning classification comprises a support vector machine (SVM).