Dependency Graphs for Automated Assistant Dialog Parameter Collection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automated assistants face challenges in efficiently modeling conversations to collect required input parameters for actions associated with third-party computing services, leading to complex models that occupy significant memory, require substantial development overhead, and result in inflexible and robotic interactions.
Innovation Solution
The use of dependency graph data structures to guide human-to-computer dialog sessions, allowing automated assistants to determine and obtain necessary parameters efficiently while optimizing hardware resource usage on user devices, such as microphones and speakers, by generating natural language outputs and receiving free-form natural language inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional automated assistants use complex conversation models to handle all possible inputs and combinations, then they can provide comprehensive service coverage, but they occupy large memory space and require substantial development overhead
Solution Approach 1:
The patent segments the conversation model into multiple specialized sub-models, each handling specific input types or combinations. Instead of one monolithic model covering all possibilities, the system divides functionality across smaller models that can be independently managed and optimized, reducing overall memory requirements while maintaining comprehensive service coverage.
Solution Approach 2:
The patent introduces a new dimension of organization by structuring models hierarchically across multiple levels or layers. Rather than a single flat model attempting to handle all inputs, the system organizes sub-models in a dimensional structure where different layers handle different aspects of conversation, enabling comprehensive coverage with reduced complexity at each individual level.
2Adaptability or versatility
If conventional automated assistants use complex conversation models to handle all possible inputs, then they can provide comprehensive service coverage, but they require updating over time to handle previously unforeseen inputs
Solution Approach 1:
By segmenting the model into specialized sub-models, the patent enables independent updates and maintenance of each segment. When new input types emerge, only the relevant sub-model needs updating rather than the entire system, significantly reducing maintenance effort while preserving comprehensive service coverage.
Solution Approach 2:
The patent implements dynamic model selection and composition, allowing the system to adapt to new inputs by activating or combining different sub-models as needed. This dynamic approach enables the system to handle previously unforeseen inputs without requiring complete model retraining, reducing maintenance burden while maintaining adaptability.
3Reliability
If automated assistants conduct extensive dialog sessions to collect all required parameters, then they can ensure complete information gathering, but they increase power consumption on user devices
Solution Approach 1:
The patent applies partial action by collecting only the essential parameters needed to fulfill user intent, rather than exhaustively gathering all possible information. The system identifies and collects the minimum necessary parameters through streamlined dialog sessions, ensuring sufficient information completeness while significantly reducing power consumption compared to comprehensive information gathering approaches.
Data Source
AI summary
Conversations are modeled using dependency graph data structures to facilitate the interaction of users with automated assistants when performing actions performed by computing services. An automated assistant may utilize a dependency graph data structure to guide or otherwise control a human-to-computer dialog session with a user, e.g., by generating one or more outputs or prompts that are presented to the user on a computing device operated by that user, and may thereby enable efficient use of technical hardware.