Natural Language Understanding via Human-Computer Collaborative Dialog
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computer systems lack the ability to engage in deep natural language understanding, failing to autonomously read, build, and communicate logical explanations of arbitrary text due to pervasive ambiguity in language, implicit meaning, and the need for human interaction to discern context.
Innovation Solution
An architecture and process that enables computers to learn and understand natural language through collaborative dialog with humans, using linguistic analysis, semantic representation, and knowledge integration to iteratively refine their understanding of texts, allowing them to answer questions and explain their reasoning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If massive data and machine learning techniques are applied to automatic question answering, then system capability is improved, but deep understanding and logical explanation remain lacking
Solution Approach 1:
The patent introduces an intermediate knowledge representation layer that bridges raw text data and question answering outputs. This intermediary structure enables the system to maintain logical understanding while achieving productive question answering through structured knowledge graphs and semantic representations.
Solution Approach 2:
The system segments the question answering process into distinct components: text ingestion, knowledge extraction, knowledge representation, and answer generation. This segmentation allows each component to specialize in maintaining logical understanding while collectively achieving high productivity.
2Speed
If shallow linguistic techniques are used, then processing speed is improved, but deep understanding of text is lost
Solution Approach 1:
The system performs preliminary deep semantic analysis and knowledge extraction during the text ingestion phase, transforming raw text into structured knowledge representations before question answering occurs. This preliminary action enables fast retrieval and processing during actual Q&A operations.
Solution Approach 2:
The patent transforms text from a linear sequence of words into a multi-dimensional knowledge structure with semantic relationships, entities, and concepts. This dimensional transformation allows the system to maintain deep understanding while enabling efficient multi-path query processing.
3Extent of automation
If autonomous text understanding is implemented without human interaction, then automation is improved, but accuracy in discerning implicit meaning deteriorates
Solution Approach 1:
The system implements feedback loops where human users can correct and refine the system's understanding of implicit meanings. This feedback mechanism allows the autonomous system to progressively improve its precision in detecting subtle semantic nuances while maintaining high automation.
Solution Approach 2:
The system autonomously identifies areas of uncertainty in implicit meaning detection and automatically seeks clarification through structured human interaction, rather than requiring constant human oversight. This self-service approach maintains automation while improving precision.
Data Source
AI summary
An architecture and processes enable computer learning and developing an understanding of arbitrary natural language text through collaboration with humans in the context of joint problem solving. The architecture ingests the text and then syntactically and semantically processes the text to infer an initial understanding of the text. The initial understanding is captured in a story model of semantic and frame structures. The story model is then tested through computer generated questions that are posed to humans through interactive dialog sessions. The knowledge gleaned from the humans is used to update the story model as well as the computing system's current world model of understanding. The process is repeated for multiple stories over time, enabling the computing system to grow in knowledge and thereby understand stories of increasingly higher reading comprehension levels.