Document Classification Using Topic and Query Rules
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing document classification methods often struggle to accurately categorize electronic documents based on their content, particularly when dealing with diverse document types such as texts, images, and music, and fail to adapt to user-defined categories effectively.
Innovation Solution
The method involves generating statistical data from training documents to create topic model-based and query-based classification rules. These rules are used to classify electronic documents, with the topic model-based rules focusing on relevant topics and the query-based rules incorporating user-defined categories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional classification methods (Naive Bayesian, support vector machines, decision trees) are used, then the classification process is simple and fast, but the classification accuracy and relevance to user-defined categories deteriorate
Solution Approach 1:
The patent segments the classification system into multiple independent rule types (topic model-based rules, query-based rules, and user-defined category rules). Each rule type handles specific aspects of document classification independently, allowing the system to achieve high accuracy through coordinated use of multiple specialized rules rather than relying on a single complex algorithm.
Solution Approach 2:
The patent introduces statistical data as an intermediary element that connects training documents to classification rules. The statistical data serves as a mediator that captures patterns from training data and translates them into actionable classification criteria, enabling the system to adapt to user-defined categories without requiring direct programming of complex classification logic.
2Adaptability or versatility
If content-based classification is used, then the classification is automated, but the adaptability to user-defined categories deteriorates
Solution Approach 1:
The patent implements dynamic classification rules that can be created and modified based on user-defined categories. The system allows users to dynamically add, remove, or modify category definitions and associated rules without requiring system reconfiguration or complex programming changes. This dynamic approach enables the classification system to adapt flexibly to changing user needs and category structures.
3Reliability
If multiple classification rules are created, then the classification relevance improves, but the processing time and computational resources increase
Solution Approach 1:
The patent applies partial action by implementing a hierarchical rule evaluation process where the system evaluates rules in a specific order and stops when a matching rule is found. Not all classification rules are applied to every document - instead, the system uses statistical data and rule priorities to determine which subset of rules is relevant for each document, reducing unnecessary computational overhead while maintaining classification relevance.
Data Source
AI summary
A method of classifying an electronic document from an electronic media source includes generating statistical data from one or more training documents and creating a plurality of classification rules, including creating at least one topic model-based classification rule using the statistical data and creating at least one query-based classification rule using one or more user defined categories and the statistical data. The method further includes classifying the electronic document using the at least one topic model-based classification rule. Example systems for classifying an electronic document from an electronic media source are also disclosed.