Keyword Set Generation for Accurate Text Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The accuracy of text classification using keywords is highly dependent on the appropriateness of the selected keywords, leading to inconsistent classification results.
Innovation Solution
An information processing device and method that includes a class acquisition unit, a candidate set generation unit, and a keyword set generation unit to determine appropriate keywords for text classification by optimizing keyword sets based on relationships between keywords, considering inter-class exclusivity and intra-class diversity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If keywords are selected for text classification, then classification can be performed, but classification accuracy varies greatly depending on keyword appropriateness
Solution Approach 1:
The system performs preliminary actions by automatically generating keyword candidate sets for each class before actual classification. It pre-determines appropriate keywords by analyzing relationships between keywords (inter-class exclusivity and intra-class diversity) so that when classification is needed, ready-to-use accurate keyword sets are available, eliminating the need for manual keyword selection and ensuring consistent high accuracy.
Solution Approach 2:
The system makes itself self-service by automatically determining appropriate keywords without human intervention. The keyword set generation unit autonomously selects keywords based on predefined criteria (inter-class exclusivity and intra-class diversity metrics), allowing the system to serve itself in keyword selection rather than requiring external manual input, thereby ensuring objective and consistent keyword choice.
2Measurement precision
If keyword sets are generated without considering relationships between keywords, then generation is simple, but classification precision is reduced due to keyword overlap
Solution Approach 1:
The system changes parameters by introducing quantitative metrics to evaluate keyword relationships. Specifically, it uses inter-class exclusivity (measuring keyword distinction between classes) and intra-class diversity (measuring keyword coverage within classes) as adjustable parameters. By optimizing keyword selection based on these parameter changes, the system achieves high classification precision while managing complexity through automated parameter-based evaluation.
Solution Approach 2:
The system implements feedback by continuously evaluating keyword relationships and using this information to refine keyword set generation. The keyword set generation unit receives feedback from relationship analysis between keywords and adjusts its selection accordingly, creating a closed-loop system that improves precision by learning from keyword interactions and optimizing future selections.
Data Source
AI summary
In an information processing device, the class acquisition means acquires multiple classes to which target data is classified. The candidate set generation means generates a keyword candidate set which is a set of keyword candidates corresponding to each class. The keyword set generation means generates a keyword set corresponding to each class, from the keyword candidate set corresponding to each class, based on a relationship between the keywords included in the keyword candidate sets using a learned optimization model for decision making in text classification.