Graph-Based Peer Detection for Patent Portfolios
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for searching and filtering large databases of patent documents, such as those used in the financial and legal sectors, often fail to adequately focus on key information, leading to inefficient and inconsistent results, as they do not effectively utilize hierarchical structures and weighted metrics to determine similarity between documents.
Innovation Solution
A graph-based system that processes search terms and applies syntax across document databases to identify peer matches by using weighted sets of classification codes defined on hierarchical taxonomy trees, allowing for a more accurate and scalable comparison of patent portfolios and other entities, enabling the identification of similar companies or objects within complex data sets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional search methods are used to search large databases of patent documents, then the search can be performed with simple algorithms, but the results do not adequately focus on key information and are inconsistent
Solution Approach 1:
The patent applies parameter changes by transforming the search problem into a graph-based similarity measurement problem. Instead of using traditional text matching parameters, the system uses graph distance metrics and hierarchical taxonomy weights to quantify similarity between patent portfolios, thereby improving search accuracy through changed measurement parameters
Solution Approach 2:
The patent introduces an intermediary graph-based representation layer between the raw patent data and the search results. This intermediary representation transforms complex patent portfolio comparisons into graph distance calculations, mediating between the raw data and meaningful similarity measurements
2Measurement precision
If hierarchical taxonomy trees with weighted classification codes are used to compare patent portfolios, then similarity measurement accuracy is improved, but computation complexity increases
Solution Approach 1:
The patent applies segmentation by breaking down the complex similarity measurement problem into hierarchical segments based on taxonomy levels. The graph-based similarity calculation is performed segment by segment through the hierarchical taxonomy structure, allowing accurate measurement while managing computational complexity through structured decomposition
Solution Approach 2:
The patent transitions from traditional flat text-based search to a multi-dimensional graph-based representation. By mapping patent portfolios onto graph structures with hierarchical taxonomy dimensions, the system achieves more accurate similarity measurements through added structural dimensions
3Ease of operation
If graph-based representation is used to represent complex data and processing results, then user understanding and visualization are improved, but system complexity increases
Solution Approach 1:
The patent creates a graphical copy or representation of the complex patent portfolio similarity data. Instead of directly presenting raw similarity scores and complex relationships, the system generates graphical visualizations that copy the essential structure and relationships in an easily interpretable format for users
Data Source
AI summary
The present invention provides a method and system delivering graph-based metric to measure a similarity between weighted sets of classifications codes (presented as nodes) defined on hierarchical taxonomy trees. The suggested method is applied to find company peers in a particular domain, e.g., the IP domain based on a company patent portfolio. The suggested method may be applied to other domains that include hierarchical classifications such as trademarks, legal documents, scientific papers, lawsuits etc.