Virtual Agent Bidding Model for Unstructured Conversations

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current chatbots are limited in handling unstructured conversations and switching between tasks, often making mistakes when deciding to switch topics, and struggle to resolve user requests that fall under different task categories due to ambiguous vocabularies and intents.

Innovation Solution

A virtual agent system employing a bidding and auction model, where multiple bidding modules generate bids based on user inputs, assign intent confidence scores, and an auctioneer module determines a winning bid to provide accurate and relevant responses, enabling efficient handling of unstructured conversations and multiple task categories.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If chat bots rely on vocabulary building to handle unstructured conversations, then entertainment values are offered, but the ability to resolve user requests across different task categories remains insufficient due to ambiguous vocabularies and intents

Engineering Contradiction:

Improveability to handle unstructured conversationsVSAvoidaccuracy in resolving user requests

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The system segments the conversation handling into multiple independent bidding modules, each specialized in specific task categories. Instead of relying on a single vocabulary-based approach, the system divides the problem into multiple specialized components that can independently evaluate and respond to user inputs, thereby improving both adaptability and reliability.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system changes the parameter of intent representation from ambiguous vocabulary matching to structured intent detection with confidence scores. By transforming the response mechanism from simple vocabulary matching to a bidding system with quantifiable confidence levels, the system achieves more reliable resolution of user requests while maintaining versatility.

Inventive Principle:

Principle #35Parameter changes

2Ease of operation

If chat bots are designed to handle a single task at a time, then task execution is straightforward, but the capability to handle unstructured conversations or topic switching is lost

Engineering Contradiction:

Improvesimplicity of task executionVSAvoidability to handle topic switching

Core Design Contradiction:

Ease of operationVSAdaptability or versatility

Solution Approach 1:

The system implements multi-functionality by designing bidding modules that can handle multiple task categories. Each module is capable of evaluating various types of user requests and competing for the winning bid, allowing the system to seamlessly handle topic switching while maintaining the simplicity of individual task execution through the auction mechanism.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The auction mechanism serves as an intermediary that coordinates between multiple specialized bidding modules. This mediator enables topic switching by selecting the most appropriate module for each user input based on confidence scores, thereby providing adaptability without compromising the ease of operation of individual tasks.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Adaptability or versatility

If chat bots use vocabulary building approaches, then entertainment responses are generated, but the system becomes complex and difficult to train

Engineering Contradiction:

Improveresponse capabilityVSAvoidtraining complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

Instead of training a single complex vocabulary-based system, the approach copies the functionality across multiple specialized bidding modules. Each module can be trained independently on specific task categories, reducing the overall training complexity while maintaining versatile response capability through the collection of specialized modules.

Inventive Principle:

Principle #26Copying

4Adaptability or versatility

If chat bots make decisions to switch topics, then conversation flexibility is improved, but mistakes are made in determining appropriateness of topic switching

Engineering Contradiction:

Improveconversation flexibilityVSAvoidaccuracy of topic switching decisions

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The system implements feedback through confidence scores that indicate the appropriateness of topic switching decisions. Each bidding module provides a confidence score reflecting its certainty about handling the user's intent, allowing the system to make reliable decisions about topic switching based on quantitative feedback rather than ambiguous vocabulary matching.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS11037048B2Virtual conversation method or system

Publication Date: 2021.06.15 SERVICENOW INC

AI summary

An automated conversation is facilitated between a user and a virtual agent. A system receives an input message from the user and analyzes an intent of the input message. Based on the intent of the input message, the system generates a plurality of bids for responding to the input message, and assigns an intent confidence score to each bid from the plurality of bids based on a confidence level of each bid from the plurality of bids. The system determines a winning bid from the plurality of bids based on the intent confidence score associated with each bid from the plurality of bids, and generates a response based on the winning bid.