Chat Reference Identification via Semantic Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing message management systems struggle to clearly display reference relationships between new and past utterances in chat conversations, leading to ambiguity and difficulty in understanding the conversation flow, especially in multi-user or fast-paced scenarios.
Innovation Solution
An apparatus and method that acquire and store messages, generating reference information based on analysis to identify and display the correct reference relationship between new and past messages, using techniques such as indirect input methods, utterance number referencing, and direct input to embed and display reference information, ensuring clarity without requiring excessive user effort.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual identifier addition is used to specify past utterances, then reference relationships can be identified, but user time and effort increase significantly
Solution Approach 1:
The system automatically analyzes the input message and identifies referenced utterances without requiring manual user input. The server performs text analysis, keyword matching, and semantic comparison to self-determine reference relationships, eliminating the need for users to manually add identifiers while maintaining high accuracy in reference identification.
Solution Approach 2:
The system pre-processes and stores historical utterance data with key features extracted before they are needed for reference identification. By preparing the data in advance with relevant keywords and semantic representations, the system can quickly match new messages against historical data without time-consuming analysis at the moment of input.
2Ease of manufacture
If simple time gap-based grouping is used for utterances, then processing is simple, but accuracy in identifying related utterances decreases when unrelated utterances cut into related ones
Solution Approach 1:
The system uses feedback from text analysis results to continuously refine utterance grouping. By analyzing keywords, semantic similarity, and contextual relationships between utterances, the system adjusts grouping decisions dynamically, correcting errors that would occur with simple time-gap-based methods and ensuring that unrelated utterances are properly separated.
Solution Approach 2:
The system changes the parameters used for grouping from purely temporal (time gaps) to include semantic parameters such as keyword similarity, topic coherence, and contextual relevance. This multi-parameter approach allows accurate distinction between related and unrelated utterances even when they occur in close temporal proximity.
3Loss of information
If expressions like 'Everything is going well>preparation' are manually input to reference past utterances, then ambiguity is reduced, but input complexity and time consumption increase
Solution Approach 1:
The system automatically generates reference expressions by analyzing the input message and matching it with historical utterances. Instead of requiring users to manually type expressions like '>preparation' or '>Mr. A', the server performs text analysis, identifies the intended reference target, and automatically inserts appropriate reference markers, thereby reducing input complexity while maintaining information clarity.
Solution Approach 2:
The system replaces manual mechanical input operations with automated computational analysis. Text analysis algorithms, keyword matching, and semantic comparison mechanisms substitute for the manual process of typing reference expressions, eliminating the need for complex user input while preserving the ability to clearly identify reference relationships.
Data Source
AI summary
In a client, an input receiver receives utterance data inputted through the client and utterance data inputted through a different client and received by a receiver, and then stores the utterance data in a past log storage as a past log. When the input receiver receives new utterance data, a reference information generator determines preceding utterance data referred to by the new utterance data, from the past log, generates reference information indicating a reference relationship, and then stores the reference information in a reference information storage. Then, a transmitter transmits the new utterance data and the reference information to the different client, and a display controller displays the utterance data stored as the past log, the new utterance data, and an indicator based on the reference information.