KPI Extraction From Unstructured Text Using Noun-Value Linking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for extracting key performance indicators (KPIs) from textual data are limited by domain-specific approaches, requiring significant resources for development and training, and struggle with unstructured data and varying language formats.
Innovation Solution
A domain-agnostic, vocabulary-independent model using entity recognition and relation extraction to extract and link KPIs, employing a segmentation model, part-of-speech tags, and classification models to identify and associate candidate nouns and values, while filtering redundant indicators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If domain-specific methods are used to extract KPIs, then extraction accuracy for specific domains is improved, but development resources and training time increase significantly
Solution Approach 1:
The patent applies universality by creating a domain-agnostic KPI extraction framework that can process text documents from any domain without requiring domain-specific training. The system uses general-purpose NLP techniques including entity recognition, relation extraction, and natural language inference that work across diverse domains such as finance, healthcare, and retail, eliminating the need for separate training processes for each domain while maintaining extraction accuracy
2Measurement precision
If domain-specific methods are used to extract KPIs, then extraction accuracy for specific domains is improved, but development resources increase significantly
Solution Approach 1:
The system employs a universal extraction framework that handles multiple domains through a single unified model, reducing development resources by eliminating the need to build and maintain separate domain-specific extraction systems. The framework processes various KPI types across different domains using the same core algorithms and data structures
Solution Approach 2:
The patent segments the KPI extraction process into distinct modular components: entity recognition module, relation extraction module, and natural language inference module. This segmentation allows each component to be developed and optimized independently, reducing overall system complexity and development resources while maintaining high extraction accuracy through specialized processing at each stage
3Measurement precision
If unstructured data with differing templates and formats is processed using non-generalized methods, then extraction precision for specific formats is improved, but adaptability to new domains decreases
Solution Approach 1:
The patent implements dynamics by creating an adaptive extraction system that automatically adjusts to different document formats, templates, and domains without requiring reconfiguration. The system dynamically identifies and processes various KPI representations through flexible entity recognition and relation extraction mechanisms, maintaining high precision across diverse formats while being easily adaptable to new domains through natural language inference
Data Source
AI summary
A method includes: determining a first bounding box in textual data; categorizing the first bounding box into a paragraph category among a list of categories including a header category, a section header category, the paragraph category, and a noise category; extracting a candidate noun from the text in the categorized first bounding box, as a candidate paragraph noun; extracting a candidate value from the text in the categorized first bounding box; generating a relationship between the candidate paragraph noun and the extracted candidate value; associating the candidate paragraph noun with a candidate header noun; and generating a key performance indicator using the candidate header noun, the candidate paragraph noun, the candidate value, and the generated relationship.