User behavior prediction method and apparatus, and behavior prediction model training method and apparatus

A user behavior prediction apparatus, wherein the apparatus comprises:

• at least one processor;
• and
• a non-transitory computer-readable storage medium coupled to the at least one processor and storing programming instructions for execution by the at least one processor, wherein the programming instructions instruct the at least one processor to perform the following operations:
• obtaining behavior prediction information that is used to predict a specified behavior, wherein the behavior prediction information comprises a plurality of pieces of characteristic data, and wherein any two pieces of the characteristic data belong to different categories;
• obtaining a first contribution value of each of the plurality of pieces of characteristic data for the specified behavior, wherein the first contribution value is used to indicate a degree of impact on execution of the specified behavior, and wherein the first contribution value is positively correlated with the degree of impact;
• processing every N pieces of the characteristic data in the plurality of pieces of characteristic data by using one corresponding characteristic interaction model, to obtain a second contribution value of the every N pieces of the characteristic data for the specified behavior, wherein N is an integer greater than 1, wherein one characteristic interaction model corresponding to any N pieces of the characteristic data is determined based on N categories to which the any N pieces of the characteristic data belong, wherein the second contribution value is used to indicate a degree of impact on execution of the specified behavior, and wherein the second contribution value is positively correlated with the degree of impact;
• and
• determining an execution probability of the specified behavior based on the obtained first contribution value of each piece of the characteristic data and the obtained second contribution value of the every N pieces of the characteristic data.

A behavior prediction model training apparatus, wherein a behavior prediction model comprises a plurality of characteristic interaction models, wherein each characteristic interaction model corresponds to N categories, wherein any two characteristic interaction models correspond to different categories, wherein the category is a category of sample characteristic data in training sample data, and wherein the apparatus comprises:

• at least one processor;
• and
• a non-transitory computer-readable storage medium coupled to the at least one processor and storing programming instructions for execution by the at least one processor, wherein the programming instructions instruct the at least one processor to perform the following operations:
• obtaining the training sample data, wherein the training sample data comprises a plurality of pieces of sample characteristic data and a behavior label of a sample behavior, wherein any two pieces of sample characteristic data belong to different categories, and wherein the behavior label is used to indicate whether a user executes the sample behavior;
• obtaining a first reference contribution value of each of the plurality of pieces of sample characteristic data for the sample behavior, wherein the first reference contribution value is used to indicate a degree of impact on execution of the sample behavior, and wherein the first contribution value is positively correlated with the degree of impact;
• processing every N pieces of the sample characteristic data in the plurality of pieces of sample characteristic data by using one corresponding characteristic interaction model, to obtain a second reference contribution value of the every N pieces of the sample characteristic data for the sample behavior, wherein the second reference contribution value is used to indicate a degree of impact on execution of the sample behavior, and wherein the second reference contribution value is positively correlated with the degree of impact;
• determining an execution probability of the sample behavior based on the obtained first reference contribution value of each piece of the sample characteristic data and the obtained second reference contribution value of the every N pieces of the sample characteristic data;
• and
• adjusting a model parameter of the behavior prediction model based on a difference between the execution probability of the sample behavior and the behavior label, and continue training until a training stop condition is met, to obtain the behavior prediction model whose model parameter is adjusted.

A non-transitory computer readable storage medium, wherein the computer readable storage medium stores an instruction, and wherein when the computer readable storage medium runs on a computer, the computer is enabled to perform:

• obtaining behavior prediction information that is used to predict a specified behavior, wherein the behavior prediction information comprises a plurality of pieces of characteristic data, and wherein any two pieces of characteristic data belong to different categories;
• obtaining a first contribution value of each of the plurality of pieces of characteristic data for the specified behavior, wherein the first contribution value is used to indicate a degree of impact on execution of the specified behavior, and wherein the first contribution value is positively correlated with the degree of impact;
• processing every N pieces of the characteristic data in the plurality of pieces of characteristic data by using one corresponding characteristic interaction model, to obtain a second contribution value of the every N pieces of the characteristic data for the specified behavior, wherein N is an integer greater than 1, wherein one characteristic interaction model corresponding to any N pieces of the characteristic data is determined based on N categories to which the any N pieces of the characteristic data belong, wherein the second contribution value is used to indicate a degree of impact on execution of the specified behavior, and wherein the second contribution value is positively correlated with the degree of impact;
• and
• determining an execution probability of the specified behavior based on the obtained first contribution value of each piece of the characteristic data and the obtained second contribution value of the every N pieces of the characteristic data.