Conversational Bot Handling Novel Utterances

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conversational marketing systems face challenges in providing real-time, coherent responses to user inputs in multi-turn conversations, as existing AI-based chatbots struggle with context switching and flexibility, especially when dealing with novel or unrecognizable user utterances.

Innovation Solution

A method that configures a conversational bot using a machine learning model trained to classify utterances into topics, employs a clustering algorithm like affinity propagation to group unrecognized utterances, and updates the model based on identified topics, while using a directed acyclic graph (DAG) to guide conversation flow and enable proactive and reactive behavior.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If a rules-based chatbot with decision trees is used, then the bot is easy to build and control proactively, but it is brittle and does not scale

Engineering Contradiction:

Improveease of buildingVSAvoidscalability

Core Design Contradiction:

Ease of manufactureVSAdaptability or versatility

Solution Approach 1:

The patent replaces the mechanical decision tree system with an AI-based intent classification system that uses machine learning to automatically understand and categorize user inputs, enabling the bot to scale without proportional increases in manual configuration effort

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables self-service through automated intent classification and context management, where the AI model autonomously determines user intent and manages conversation state without requiring manual rule updates for each scenario

Inventive Principle:

Principle #25Self-service

2Reliability

If an intent-based bot is used, then the bot is more robust, but it is not flexible to context switches once an intent is activated

Engineering Contradiction:

ImproverobustnessVSAvoidcontext switching flexibility

Core Design Contradiction:

ReliabilityVSAdaptability or versatility

Solution Approach 1:

The patent implements dynamic context management where the system can transition between different intent states based on user inputs, allowing flexible context switching while maintaining the robustness of structured intent classification through stateful conversation management

Inventive Principle:

Principle #15Dynamics

3Adaptability or versatility

If a machine learning model is used to classify utterances, then the bot can handle novel utterances, but the system complexity increases

Engineering Contradiction:

Improvehandling novel utterancesVSAvoidsystem complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary layer of automated intent classification that bridges raw user inputs and bot responses, using machine learning models to handle novelty while managing complexity through standardized classification pipelines and pre-defined intent schemas

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS12057106B2Authoring content for a conversational bot

Publication Date: 2024.08.06 SALESLOFT INC

AI summary

A method of authorizing content for use, e.g., in association with a conversational bot. The method begins by configuring a conversational bot using a machine learning model trained to classify utterances into topics. Utterances that are not recognized by the machine learning model (e.g., according to some configurable threshold) are then identified. Using a clustering algorithm, one or more of the identified utterances are then processed into a grouping. Information identifying a topic associated with the grouping is then received and, in response, the machine learning model is updated to include the topic.