Conversation Assistance Using Term Groups for Real-Time Remark Evaluation

The patent segments the conversation evaluation process into distinct modules: voice analysis for extracting speech content, word group analysis for categorizing terms into semantic groups (greetings, counterarguments, descriptions, suspicions, excuses, agreements, promises, affirmations), and remark evaluation for assessing conversation quality. This segmentation enables real-time processing by analyzing smaller units sequentially rather than evaluating entire conversations at once, thus resolving the contradiction between evaluation accuracy and real-time speed.

The system performs preliminary actions by pre-defining term groups and their categories before conversation evaluation begins. The voice analysis and word group analysis components are prepared in advance with predefined criteria, allowing the remark evaluation to proceed rapidly during actual conversation without delayed processing, thereby achieving real-time evaluation while maintaining accuracy.

The patent implements a universal term group analysis framework that handles multiple conversation aspects (greetings, counterarguments, descriptions, suspicions, excuses, agreements, promises, affirmations) through a single integrated system. The same analytical infrastructure processes all term groups uniformly, providing comprehensive conversational effectiveness evaluation without proportionally increasing system complexity. This multi-functional approach allows the system to evaluate various conversational dimensions using consistent methods.

The system uses predefined term group templates that can be copied and applied repeatedly during conversation analysis. Instead of creating complex analysis logic for each conversation type, the system copies and adapts established term group categories and evaluation criteria, simplifying the system architecture while maintaining comprehensive evaluation capabilities across different conversational contexts.