Time series pattern generating system and method using historical information

A time series pattern generating system for analyzing history information generated upon execution of a business process constituted by one or more operations and defining a series of processes, and generating a time series pattern, comprising:

• a storage system which stores the history information in chronological order, the history information including, for each operation of each instance of the business process, a business process identifier, an operation identifier, an instance identifier, and information indicating a processing time of an operation of the one or more operations identified by the operation identifier;
• and
• a computer which generates the time series pattern on the basis of the history information,
• the computer comprising a processor, an input module into which the history information is input, and an output module which outputs the time series pattern, and a memory having computer-readable instructions stored thereon for execution by the processor to implement a device for generating the time series pattern,
• wherein the device is configured to:
• extract a set of history information from the history information based on common instance identifiers for respective instances and generate instance history information;
• create processing order information relating to the one or more operations of the respective instances on the basis of the generated instance history information;
• estimate a dependence relationship between the one or more operations on the basis of the created processing order information, express the dependence relationship using an adjacency matrix that represents each of the one or more operations using a respective row and a respective column and indicates each dependence between the one or more operations using a respective entry of the adjacency matrix, minimize a number of dependencies indicated in the adjacency matrix to obtain a processing order of the one or more operations by removing indications of redundant dependencies from the adjacency matrix, and generate the time series pattern including the processing order of the one or more operations;
• and
• update the adjacency matrix expressing the estimated dependence relationship by calculating a difference between a row representing a first component having a first processing order and a row representing a second component having a second processing order that is later than the first processing order.

A time series pattern generating method for analyzing history information generated upon execution of a business process constituted by one or more operations and defining a series of processes, and generating a time series pattern,

• the time series pattern generating method being implemented by a computer comprising a processor and a memory having computer-readable instructions stored thereon for execution by the processor to implement the method,
• the history information comprising records each including a business process identifier, an operation identifier, an instance identifier, and information indicating a processing time of an operation of the one or more operations identified by the operation identifier,
• the time series pattern generating method comprising:
• generating a record in the history information every time the one or more operations are executed;
• storing the generated history information in the storage system in chronological order;
• extracting a set of history information records from the history information based on common instance identifiers for respective instances of the business process and generating instance history information;
• creating processing order information relating to the one or more operations of the respective instances on the basis of the generated instance history information;
• estimating a dependence relationship between the one or more operations on the basis of the created processing order information, expressing the dependence relationship using an adjacency matrix that represents each of the one or more operations using a respective row and a respective column and indicates each dependence between the one or more operations using a respective entry of the adjacency matrix, minimizing a number of dependencies indicated in the adjacency matrix to obtain a processing order of the one or more operations by removing indications of redundant dependencies from the adjacency matrix, and generating the time series pattern including the processing order of the one or more operations;
• and
• updating the adjacency matrix expressing the estimated dependence relationship by calculating a difference between a row representing a first component having a first processing order and a row representing a second component having a second processing order that is later than the first processing order.