Content-First Interaction Model for Voice Apps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current interaction application development processes require manual creation and customization of interaction models, endpoint business logic, and content, which are time-consuming and inefficient, especially for handling multiple intents and platforms, and do not support flexible response generation without exact matching of utterances.
Innovation Solution
A content-first development platform with a pre-populated general interaction model that uses graph traversal and content index searching to determine responses, allowing for abstract intents and open-ended slots, enabling flexible and platform-agnostic development without manual coding, and supports SSML processing for human-like voice responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If manual creation and customization of interaction models, endpoint business logic, and content is used, then the application can be tailored to specific enterprise needs, but the development process becomes time-consuming and inefficient
Solution Approach 1:
The patent applies preliminary action by providing a library of pre-built interaction models, endpoint business logic templates, and content templates that can be selected and customized for different enterprise needs. This allows developers to start with pre-prepared components rather than creating everything from scratch, significantly reducing development time while maintaining adaptability through customization options.
Solution Approach 2:
The patent implements universality by designing interaction models and business logic templates that can serve multiple enterprise scenarios and platforms. A single interaction model can be adapted to work across different assistant platforms (Alexa, Google Assistant, etc.), and templates can be reused across multiple applications, reducing redundant development work while maintaining specific enterprise requirements.
2Manufacturing precision
If exact matching of utterances to intents is required, then the interaction model can be precisely controlled, but it cannot handle flexible response generation for varied user requests
Solution Approach 1:
The patent applies parameter changes by introducing confidence scores and similarity thresholds that allow the system to handle partial matches and variations in user utterances. Instead of requiring exact string matching, the system evaluates the degree of match between utterances and sample utterances, enabling flexible response generation while maintaining controlled intent recognition through adjustable matching parameters.
Solution Approach 2:
The patent introduces an intermediary layer between utterance input and intent determination that processes and evaluates multiple potential matches. This intermediary mechanism allows the system to handle ambiguous or varied user requests by selecting the most appropriate intent based on confidence scores, sample utterance similarity, and contextual factors, thereby bridging the gap between precise control and flexible response.
3Adaptability or versatility
If custom development processes are used for each interaction application, then the application can be highly specialized, but the process cannot be easily replicated across multiple platforms
Solution Approach 1:
The patent implements universality by creating platform-agnostic interaction models and business logic templates that can be deployed across multiple assistant platforms (Amazon Alexa, Google Assistant, Microsoft Cortana, etc.). The same interaction model can target different platforms without requiring complete redesign, reducing development process complexity while maintaining platform-specific optimizations through configuration rather than structural changes.
Solution Approach 2:
The patent applies segmentation by separating the interaction model into platform-independent components (intents, slots, sample utterances, business logic) and platform-specific implementation details. This segmentation allows the core interaction logic to be developed once and reused across platforms, while only the platform-specific adaptation layer needs to be customized, thereby reducing overall development complexity.
4Reliability
If comprehensive sample utterances are manually entered for each intent, then the interaction model can accurately cover all expected user inputs, but the development and maintenance becomes extremely time-consuming
Solution Approach 1:
The patent applies preliminary action by providing pre-populated libraries of sample utterances for common intents that can be directly used or lightly customized. Instead of requiring developers to manually create every sample utterance from scratch, the system provides a head start with pre-written, tested sample utterances that cover typical user inputs, significantly reducing the time and effort required for development while maintaining reliable intent recognition.
Solution Approach 2:
The patent implements self-service by enabling the system to automatically generate and suggest sample utterances based on intent definitions and contextual analysis. The system can self-populate interaction models with relevant sample utterances, reducing the manual burden on developers while ensuring comprehensive coverage of expected user inputs through automated generation and suggestion mechanisms.
Data Source
AI summary
Among other things, a developer of an interaction application for an enterprise can create items of content to be provided to an assistant platform for use in responses to requests of end-users. The developer can deploy the interaction application using defined items of content and an available general interaction model including intents and sample utterances having slots. The developer can deploy the interaction application without requiring the developer to formulate any of the intents, sample utterances, or slots of the general interaction model.