Task processing method and related product

Abstract

The invention discloses a task processing method and a related product. The method comprises the steps that first data are sorted based on time sequence characteristics, a first observation sequence corresponding to the first data is determined, and the first data comprise user behavior data or data generated in the task execution process; based on the first observation sequence and a first state probability, a first state transition probability matrix and a first observation probability matrix of a hidden Markov model, a first conditional probability or a second conditional probability is obtained through prediction, the first conditional probability is used for representing the probability of the intention of creating the task by the user, and the second conditional probability is used for representing the probability of the intention of creating the task by the user; the second conditional probability is used for representing the probability of the intention that the user continues to execute the task; and processing the task based on the first conditional probability or the second conditional probability.

Images