Time-Ordered Sequence Rule Generation via Segmented Association Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing association analysis methods are limited as they do not consider the timing data indicating the time order of data items for transactions, which is crucial for generating effective sequence rules.
Innovation Solution
A method that performs association analysis independently of timing data, extracts transaction data with timing information, and generates a time-ordered sequence rule by examining the data to identify a certain data item for each transaction, thereby specifying a time-ordered premise and predicted result.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If association analysis is performed without considering timing data, then the analysis process is simpler and faster, but the generated rules cannot accurately predict sequential transactions
Solution Approach 1:
The patent segments the rule generation process into two distinct phases: first performing association analysis without timing data to identify potential relationships, then separately analyzing timing data to determine sequential patterns. This segmentation allows each phase to optimize for its specific goal while contributing to the overall prediction accuracy.
Solution Approach 2:
The patent merges the results from association analysis (identifying item relationships) with timing data analysis (determining sequence patterns) to create comprehensive sequence rules. By combining these two analytical approaches, the system achieves both processing efficiency and prediction accuracy.
2Measurement precision
If timing data is considered in the association analysis, then the prediction accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent divides the complex analysis into manageable segments: association rule generation without timing constraints, followed by separate timing pattern analysis. This reduces computational complexity by avoiding the need to process all timing data simultaneously with association rules.
Solution Approach 2:
The patent performs preliminary association analysis to identify potential item relationships before applying timing data filters. This preliminary action narrows down the search space, making subsequent timing-based analysis more efficient and less computationally intensive.
3Reliability
If timing data is extracted and examined for each transaction, then time-ordered sequence rules can be generated, but the processing time increases
Solution Approach 1:
The patent extracts and examines timing data in advance to identify sequential patterns before final rule generation. This preliminary examination of timing data allows the system to pre-process and organize temporal information, reducing the time required for final rule creation and validation.
Solution Approach 2:
The patent extracts only the necessary timing data elements required for sequence rule generation, rather than processing all available transaction data. This selective extraction reduces processing time while maintaining the reliability needed for accurate sequential prediction.
Data Source
AI summary
Methods, computer program products, and systems are presented. The method, computer program products, and systems can include, for instance: performing an association analysis on a transaction dataset to return an association rule, wherein the association analysis is performed independent of timing data indicating a time order of data items for the respective transactions; extracting transaction data defining a set of transactions of the transaction dataset using the association rule, the transaction data including timing data indicating a time order of data items for the set of transactions; examining data of the transaction data using the timing data indicating the time order of data items for transactions of the set of transaction; returning a time ordered sequence rule specifying a time ordered premise and predicted result, wherein the time ordered sequence rule is in dependence on the examining; monitoring activities of a user for occurrence of a condition defining the premise of the sequence rule; and returning an action decision in response an occurrence of the condition.