Conversation System Using Code-Based Response Scoring

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional dialogue systems face limitations in understanding user inputs, particularly in open-ended conversations, due to their reliance on pattern matching, algorithms, or artificial neural networks, which require extensive data and resources, leading to frustration and reduced user engagement.

Innovation Solution

A conversation system that generates responses by converting user inputs into codes, using Elasticsearch and a database to score and select appropriate phrases, allowing for context-aware and emotionally nuanced interactions without the need for extensive training or data, enabling more flexible and human-like conversations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If pattern matching systems are used to handle user input, then the system can respond to predetermined patterns, but it cannot handle user input that does not exactly match the predetermined patterns

Engineering Contradiction:

Improveresponse accuracyVSAvoidhandling variety of user input

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent introduces an intermediary component that translates user input into codes representing user intent. This intermediary layer (code generation module) bridges the gap between raw user input and the pattern matching system, allowing the system to handle diverse inputs by converting them into standardized codes that can be matched against predetermined patterns.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system changes the parameter of user input representation by converting free-text input into discrete codes. This parameter transformation allows the flexible, unstructured user input to be converted into a structured format that can be reliably matched against predetermined patterns, thus maintaining both adaptability and reliability.

Inventive Principle:

Principle #35Parameter changes

2Adaptability or versatility

If algorithms are used to score user input and match with response classes, then the system can assist with pattern matching limitations, but there is no guarantee of a perfect match and responses may not make sense to the user

Engineering Contradiction:

Improveflexibility in handling user inputVSAvoidresponse accuracy

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent implements a feedback mechanism where the system generates multiple candidate responses, scores them, and selects the highest scoring response. This feedback loop ensures that the system continuously evaluates and refines its response selection based on scoring criteria, improving the reliability of responses while maintaining flexibility in handling various user inputs.

Inventive Principle:

Principle #23Feedback

3Adaptability or versatility

If artificial neural networks are trained to match user inputs to user intents, then the system can handle a variety of user inputs, but it requires a lot of data and prohibitive amounts of computer resources

Engineering Contradiction:

Improveability to match various user inputsVSAvoiddata requirements and computational resources

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent replaces the expensive, resource-intensive neural network approach with a more economical code-based system. Instead of requiring extensive training data and computational resources, the system uses lightweight code generation and matching mechanisms that achieve similar adaptability without the prohibitive costs of neural network training and deployment.

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

4Ease of operation

If conventional dialogue systems are used for conversation, then they can handle simple tasks, but they repeat themselves, stop making sense, or fall back to default responses in open-ended conversations

Engineering Contradiction:

Improveability to handle simple tasksVSAvoidconversation coherence

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The patent applies preliminary action by pre-defining multiple candidate responses for each code and pre-establishing scoring criteria. This preparation in advance allows the system to quickly and coherently select appropriate responses during conversation without repeating itself or falling back to default responses, maintaining conversation coherence even in open-ended scenarios.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11055493B2Conversation system

Publication Date: 2021.07.06 LEWIS JAMES

AI summary

A conversation system allowing for complex, open-ended conversations with users and method of its operation are disclosed. An example method includes identifying a conversation and a user, encoding the user's input message and finding codes in other texts, generating responses, and selecting an appropriate response, and sending the selected response to the user. The conversation system utilizes various methods, scoring, codes, and the like to find the appropriate responses. Example methods also include steps for sending short sentences representing action, or sending a “thought” or “daydream” to the user that reveals the emotional state of the conversation system with respect to the user, and steps for changing that emotional state. The system avoids reliance on artificial neural networks and keeps track of the context of a conversation by saving conversation codes in a database.