Competitor Identification via Document Relevance Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current search engines fail to provide a comprehensive overview of documents related to a specific topic and cannot focus searches on specific aspects of a topic, leading to relevant documents being overlooked, as they rely solely on keyword matching and lack the ability to analyze document context and relevance.
Innovation Solution
A system that identifies lines of business for companies by analyzing document relevancy using document classification models, which evaluate documents based on patterns and scores to associate business lines with companies and determine competitive relationships between them.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If search engines use keyword matching only, then the search process is simple and fast, but relevant documents are overlooked and comprehensive overview is not provided
Solution Approach 1:
The patent segments the document analysis process into multiple independent components: keyword matching module, semantic analysis module, and relevance scoring module. Each component processes different aspects of document relevance independently, allowing the system to maintain speed while comprehensively evaluating both explicit keywords and implicit semantic meaning to avoid overlooking relevant documents.
Solution Approach 2:
The patent transitions from one-dimensional keyword matching to multi-dimensional document evaluation by introducing semantic analysis dimensions, contextual relationship dimensions, and relevance scoring dimensions. This dimensional expansion enables the system to capture both explicit keyword matches and implicit semantic relevance, preventing information loss while maintaining efficient processing through parallel evaluation of multiple dimensions.
2Device complexity
If search engines rely on keyword matching, then the system complexity is low, but the ability to focus search on specific aspects is not available
Solution Approach 1:
The patent implements a universal search processing framework that handles multiple search objectives through a single integrated system. The same semantic analysis and relevance scoring mechanisms serve both broad overview searches and focused aspect searches, allowing the system to adapt to different search intents without requiring separate specialized systems for each search type.
Solution Approach 2:
The patent introduces dynamic search capability where the system can adapt its analysis depth and focus based on user input and query context. The relevance scoring module dynamically adjusts its evaluation criteria based on the search query, enabling the system to transition between comprehensive overview mode and focused aspect mode, thereby providing versatility while managing complexity through adaptive rather than static architecture.
3Measurement precision
If the system analyzes document relevancy using classification models, then the accuracy of business line identification is improved, but the processing time and computational resources increase
Solution Approach 1:
The patent applies preliminary action by pre-processing documents to extract and store key features, semantic representations, and relevance indicators before the actual classification task. The system pre-computes semantic embeddings and stores document metadata that can be quickly retrieved and matched during classification, reducing the time required for real-time analysis while maintaining high accuracy in business line identification.
Solution Approach 2:
The patent utilizes parameter changes by adjusting the complexity and depth of semantic analysis based on document characteristics and search query requirements. The relevance scoring module can modify its evaluation parameters dynamically, using simpler matching criteria for quick assessments and more comprehensive analysis only when necessary, thereby balancing processing time with identification accuracy across different document types and search contexts.
Data Source
AI summary
Some embodiments provide a method for identifying competitors of a particular company. The method identifies a set of potential competitors of a particular company. For each potential competitors, the method calculates a score quantifying the competitive relationship of the potential competitor to the particular company. When the calculated score is above a particular threshold, the method associates the potential competitor as a competitor of the particular company.