Interactive Sequential Pattern Mining System

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing sequential pattern mining algorithms, such as Apriori and pattern growth, are computationally expensive and lack user-controlled focus, returning many irrelevant patterns due to excessive database scans and limited user input, with domain experts facing challenges in specifying formal constraints and lacking interactivity in the mining process.

Innovation Solution

An interactive approach for sequential pattern mining that integrates user-defined constraints dynamically, allowing domain experts to iteratively explore and rank candidate patterns based on interestingness, using a support counting algorithm and interaction processor to guide the mining process, reducing the search space and focusing on relevant patterns.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If existing sequential pattern mining algorithms (Apriori, pattern growth) are used, then complete pattern discovery is achieved, but computational cost and time consumption increase significantly

Engineering Contradiction:

Improvepattern discovery completenessVSAvoidmining efficiency

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The system performs preliminary actions by pre-processing the sequence database to identify and store frequent itemsets before the actual pattern mining process. This preliminary computation reduces the search space for subsequent mining operations, thereby improving efficiency without sacrificing pattern discovery completeness

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The mining process is segmented into distinct phases: frequent itemset generation, candidate pattern generation, and pattern evaluation. Each phase operates independently with optimized algorithms, allowing the system to process large databases efficiently while maintaining comprehensive pattern discovery

Inventive Principle:

Principle #1Segmentation

2Reliability

If existing sequential pattern mining algorithms are used, then all frequent patterns are returned, but the number of irrelevant patterns increases

Engineering Contradiction:

Improvepattern discovery completenessVSAvoidsignal-to-noise ratio

Core Design Contradiction:

ReliabilityVSLoss of information

Solution Approach 1:

The system incorporates feedback mechanisms where mining results are evaluated against user-defined constraints and domain knowledge. Patterns that do not meet relevance criteria are filtered out, while interesting patterns are highlighted for further analysis. This feedback loop continuously refines the output quality by eliminating irrelevant patterns

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system dynamically adjusts mining parameters such as minimum support thresholds and constraint weights based on the characteristics of the discovered patterns and user preferences. This adaptive parameter adjustment optimizes the balance between pattern completeness and relevance, reducing the proportion of irrelevant patterns in the results

Inventive Principle:

Principle #35Parameter changes

3Adaptability or versatility

If user-defined constraints are integrated interactively, then domain expertise is incorporated, but system complexity increases

Engineering Contradiction:

Improvedomain knowledge integrationVSAvoidsystem architecture complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system introduces an interaction processor as an intermediary layer between the sequence database and the pattern mining algorithms. This mediator handles user-defined constraints, translates domain knowledge into mining parameters, and manages the iterative interaction process, thereby simplifying the overall system architecture while enabling sophisticated domain expertise integration

Inventive Principle:

Principle #24Intermediary (Mediator)

4Ease of operation

If the mining process is made interactive, then user control and focus are improved, but processing time for each iteration increases

Engineering Contradiction:

Improveuser control capabilityVSAvoidinteractive processing time

Core Design Contradiction:

Ease of operationVSLoss of time

Solution Approach 1:

The system performs preliminary computations including frequent itemset generation and candidate pattern identification before interactive user selection. This pre-processing reduces the computational burden during interactive iterations, allowing users to explore patterns with full control while minimizing the time penalty of interactivity

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS10565218B2Interactive sequential pattern mining

Publication Date: 2020.02.18 MICRO FOCUS LLC

AI summary

Interactive sequential pattern mining is disclosed. One example is a system including a sequence miner, and an interaction processor. A sequence database is received, the sequence database including a plurality of input sequences, where each sequence of the plurality of input sequences is an ordered list of events, and each event in the list of events includes at least one item. The sequence miner mines the sequence database for a plurality of candidate sequence patterns, the mining based on an interaction with a user. The interaction processor processes the interaction with the user, the interaction based on domain relevance of the plurality of candidate sequence patterns to the user.