Semantic Relation Extraction from Patent Hierarchies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing thesauri for semantic relation extraction, such as hypernym/hyponym relations, are not applicable to patent analysis due to technical terms and varying field-specific relations, and often lack rigid 'is-a' expressions, necessitating a novel method to extract semantic relations from patent documents with reduced manual effort.
Innovation Solution
A computer-implemented method that builds hierarchical structures from patent documents, extracts keyword lists for ancestor and descendant elements, calculates statistical indices of association using pointwise mutual information, and filters pairs to output candidate lists of keyword pairs with semantic relations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing thesauri are used for semantic relation extraction, then general semantic relations can be obtained, but they are not applicable to patent analysis due to technical terms and field-specific relations
Solution Approach 1:
The patent applies local quality by creating field-specific thesauri tailored to different technical domains (e.g., electronics, mechanical engineering, chemistry) rather than using a single general thesaurus. Each domain receives customized semantic relations that reflect its specific terminology and concepts, thereby improving adaptability to patent analysis while maintaining reliability through domain expertise.
Solution Approach 2:
The system dynamically adjusts the thesaurus parameters by selecting different thesauri based on the detected technical field of the patent document. This parameter change allows the system to switch between general and domain-specific semantic relations, resolving the contradiction between broad applicability and domain-specific accuracy.
2Reliability
If manual work is used to define semantic relations in documents, then accurate relations can be captured, but the process is time-consuming and not scalable
Solution Approach 1:
The system implements self-service by automatically extracting and organizing semantic relations from patent documents using computational methods. The automated pipeline includes document parsing, entity recognition, relation extraction, and thesaurus generation without requiring manual intervention for each document, thereby achieving both accuracy through structured processing and high productivity through automation.
Solution Approach 2:
The patent replaces the manual mechanical process of defining semantic relations with an automated computational system. Natural language processing algorithms, statistical methods, and machine learning models substitute human analysts, enabling scalable extraction of accurate semantic relations from large volumes of patent documents.
3Loss of information
If hypernym/hyponym relations are extracted from patent documents, then semantic understanding is improved, but patent-specific words and phrases may not be covered in existing thesauri
Solution Approach 1:
The system performs preliminary action by building domain-specific thesauri before analyzing patent documents. This pre-processing step involves collecting patent-specific terminology and defining appropriate hypernym/hyponym relations for each domain, ensuring that the semantic extraction process has the necessary vocabulary and relations ready to capture patent-specific concepts accurately.
Solution Approach 2:
The patent adds a new dimension to the thesaurus structure by incorporating domain-specific layers alongside general semantic relations. This multi-layered approach allows the system to simultaneously handle both general semantic concepts and patent-specific terminology, resolving the coverage gap without losing semantic information.
Data Source
AI summary
A computer-implemented method for extracting semantic relations is disclosed. In the method, a plurality of hierarchal structures that originates from a corpus of documents is obtained. Each hierarchal structure includes a plurality of elements having respective recitations included in a corresponding document. In the method, for each predetermined relationship between ancestor and descendant elements in the hierarchal structures, a first keyword list is extracted from the ancestor element and a second keyword list is extracted from the descendant element. A statistical index is calculated for each pair of first and second keywords using the first keyword lists and the second keyword lists. The index indicates a strength of association between the first and second keywords. In the method, a candidate list of keyword pairs having semantic relationships is output using the statistical index calculated for each pair.