Conversational AI Query Response Using Unified Contextual Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dialogue systems face challenges in processing mixed structured and unstructured data, leading to difficulties in identifying information and requiring extensive training resources and time for neural networks.
Innovation Solution
Systems and methods that generate contextual data using both structured and unstructured data, processed by neural networks, allowing for efficient response generation without extensive training, by converting structured data into unstructured form and utilizing unstructured training data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If both structured data and unstructured data are stored in the knowledge database, then the information coverage is improved, but it becomes difficult for the dialogue manager to identify the needed information due to different representation formats
Solution Approach 1:
The patent applies homogeneity by converting structured data into unstructured text format, making all data in the knowledge database have the same textual representation. This allows the dialogue manager to process both originally structured and originally unstructured data uniformly through text matching, eliminating the difficulty of identifying information across different data formats while maintaining comprehensive information coverage
2Reliability
If training data representing samples for each identifier in structured data is used, then the neural network can interpret requests associated with identifiers, but the computing resources and time required for training increase significantly
Solution Approach 1:
The patent extracts the structured data from its tabular format and converts it into unstructured text format for training purposes. Instead of training the neural network on structured data with identifiers and samples for each identifier, the system extracts the information and presents it as natural language text, allowing the network to learn interpretation capabilities with significantly reduced training time and computational resources
3Ease of manufacture
If structured data with fields associating specific identifiers with information is used, then the data organization is improved, but training the neural network requires large amounts of training data and computing resources
Solution Approach 1:
The patent changes the parameter of data representation from structured format (with identifiers and fields) to unstructured text format. This parameter change allows the system to maintain good data organization through textual coherence and context while dramatically reducing the quantity of training data needed, as the neural network learns from natural language patterns rather than requiring extensive sampled data for each identifier
Data Source
AI summary
In various examples, contextual data may be generated using structured and unstructured data for conversational AI systems and applications. Systems and methods are disclosed that use structured data (converted to unstructured form) and unstructured data, such as from a knowledge database(s), to generate contextual data. For instance, the contextual data may represent text (e.g., narratives), where a first portion of the text is generated using the structured data and a second portion of the text is generated using the unstructured data. The systems and methods may then use a neural network(s), such as a neural network(s) associated with a dialogue manager, to process input data representing a request (e.g., a query) and the contextual data in order to generate a response to the request. For instance, if the request includes a query for information associated with a topic, the neural network(s) may generate a response that includes the requested information.