DOM Mining with SiSTeR Edit Distance Algorithm
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing algorithms for mining HTML Document Object Model (DOM) structures face significant computational challenges in efficiently identifying patterns and subtrees within complex web applications, particularly due to the high execution time required for computing tree edit distances, which hinders interactive pattern discovery and data mining tasks.
Innovation Solution
The implementation of the Restricted Top-Down edit Metric (RTDM) and its variant, Similar Sibling Trees-aware RTDM (SiSTeR), which compute edit distances between subtrees in a bottom-up manner using dynamic programming, allowing for efficient detection of similar subtrees regardless of their quantity and structure, thereby reducing computation time from O(n^4) to O(n^2) and enabling improved performance in DOM mining tasks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If tree edit distance is computed for all subtrees using existing algorithms, then similarity measurement between DOM structures is achieved, but execution time becomes magnitudes of order higher
Solution Approach 1:
The patent segments the DOM tree into smaller subtrees and processes them hierarchically from leaves to root. By dividing the computation into manageable segments (subtrees) and processing them in a bottom-up manner, the algorithm avoids computing edit distances for all possible subtree combinations, thus reducing execution time while maintaining measurement accuracy through systematic comparison of segmented structures.
Solution Approach 2:
The patent performs preliminary actions by pre-computing edit distances between leaf nodes and storing them in a distance matrix before proceeding to higher levels of the tree hierarchy. This preliminary computation of base cases enables efficient calculation of parent node distances without re-computing from scratch, significantly reducing overall execution time while preserving accurate similarity measurements.
2Adaptability or versatility
If patterns are searched in web pages with many elements, then comprehensive pattern discovery is achieved, but computation time increases significantly
Solution Approach 1:
The patent segments the large DOM tree into smaller subtrees and processes them hierarchically. By dividing the comprehensive pattern search into segmented subtree comparisons rather than evaluating all possible patterns across the entire tree, the algorithm maintains pattern discovery comprehensiveness while reducing computation time through manageable processing units.
Solution Approach 2:
The patent introduces a hierarchical dimension to the pattern search process by organizing computations from leaf nodes upward through intermediate levels to the root. This dimensional transformation from flat exhaustive search to hierarchical structured computation enables comprehensive pattern discovery across all tree levels while improving productivity through systematic organization of the search space.
3Ease of operation
If interactive pattern discovery is enabled in large DOM structures, then user interaction is achieved, but computation time remains an issue
Solution Approach 1:
The patent performs preliminary computations by pre-calculating and storing edit distances between leaf nodes in a distance matrix before interactive queries are made. This preliminary action creates a ready-to-use foundation that enables fast responsive interaction for pattern discovery without requiring time-consuming computations during user interactions, thus maintaining ease of operation while reducing perceived computation time.
Solution Approach 2:
The patent segments the computation into hierarchical levels, allowing interactive pattern discovery to focus on specific subtree regions rather than re-computing entire tree structures. This segmentation enables responsive user interaction by limiting computation to relevant segments while maintaining the ability to perform comprehensive analysis when needed.
Data Source
AI summary
Systems and methods of mining Web applications are disclosed. An example method may include selecting a distance measure between subtrees of the Web applications. The method may also include computing an edit distance for the subtrees. The method may also include finding sets of similar subtrees using the edit distance with a Similar Sibling-Trees-aware RTDM (SiSTeR) approach.