Conversational Event Modeling for Dialog Stack Routing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conversational systems face difficulties in processing ambiguous human inputs due to the complexity of conversational structures, making it challenging to determine the appropriate level of interaction within a hierarchical dialog stack, especially for tasks like ordering a pizza where inputs like 'help' or 'cancel' are unclear in their application.

Innovation Solution

The approach involves generating and propagating events throughout a dialog stack, allowing each dialog to determine its confidence level in processing the input, and selecting the most appropriate dialog to handle the event based on confidence levels, ensuring robust handling of ambiguous inputs across the conversational system.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a hierarchical dialog stack structure is used to model complex conversations, then the system can handle multi-level dialog contexts, but it becomes difficult to determine which dialog level should process ambiguous inputs like 'help' or 'cancel'

Engineering Contradiction:

Improveability to handle multi-level dialog contextsVSAvoiddifficulty in determining appropriate processing level for ambiguous inputs

Core Design Contradiction:

Adaptability or versatilityVSDifficulty of detecting and measuring

Solution Approach 1:

The system segments the dialog processing task by assigning different levels of the dialog stack to handle different types of inputs. Each dialog level can independently evaluate its confidence in processing an input, allowing ambiguous inputs to be routed to the most appropriate processing level based on confidence scores rather than forcing a single hierarchical decision path

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system introduces an event propagation mechanism as an intermediary between input reception and dialog processing. When an input is received, events are propagated through the dialog stack, allowing each dialog level to evaluate the input and determine its confidence in processing it, thereby mediating the decision of which level should handle ambiguous inputs

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If events are propagated to all dialogs in the stack for evaluation, then the system can accurately determine the most appropriate processing dialog, but this increases computational overhead and processing time

Engineering Contradiction:

Improveaccuracy in determining the most appropriate processing dialogVSAvoidprocessing time for event propagation across multiple dialogs

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system implements partial action by allowing dialogs to stop propagating events once a sufficiently confident processor is found. Rather than requiring all dialogs in the stack to evaluate every input, the propagation can be terminated early when an appropriate dialog is identified, reducing unnecessary computational overhead while maintaining accuracy

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS11151333B2Conversational event modeling

Publication Date: 2021.10.19 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

Conversational event modeling for determining how to process an input in a dialog stack of a computer-executed conversational system. Receipt of an input at an active dialog in a dialog stack may result in generation of an event corresponding to the input. The event may be propagated through the dialog stack. Dialogs in the dialog stack may generate a confidence value in relation to processing the input, and selection of the dialog in the dialog stack for use in processing the input may be based at least in part on the confidence values. In turn, the conversational system may facilitate improved handling of ambiguous or unrelated inputs at dialogs by propagating the event associated with such an input through the dialog stack. The event creation and processing functions may be provided as parameters for dialogs in a modular dialog creation system.