Temporal Expression Parsing in Conversational Data-to-Text Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing natural language querying systems fail to accurately recognize and parse non-contiguous temporal expressions, leading to incomplete or incorrect parsing of time intervals and durations.
Innovation Solution
A conversational data-to-text system that utilizes a processor and memory configured to generate n-gram sets, traverse parse trees based on grammar templates, and generate semantic frames to accurately identify and normalize temporal expressions, even when they are not contiguous.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional temporal expression parsing methods are used, then simple contiguous temporal expressions can be recognized, but non-contiguous temporal expressions cannot be accurately parsed
Solution Approach 1:
The patent segments the temporal expression parsing process into multiple components: n-gram generation, parse tree construction, and semantic frame extraction. This segmentation allows the system to handle complex non-contiguous temporal expressions by processing them in discrete, manageable steps rather than as a single monolithic operation.
Solution Approach 2:
The patent implements nested processing structures where parse trees are built within semantic frames, and n-grams are nested within the parse tree structure. This nested approach enables the system to capture hierarchical relationships in temporal expressions, allowing inner temporal expressions to be properly identified even when embedded within larger linguistic structures.
2Productivity
If simple parsing rules are applied, then processing speed is maintained, but complex temporal expressions with interruptions are parsed incorrectly
Solution Approach 1:
The patent performs preliminary actions by generating n-grams and constructing parse trees before final semantic frame extraction. This preliminary processing prepares the data structure in advance, allowing the final interpretation step to focus solely on accurate temporal expression identification without re-processing the entire input, thus maintaining speed while improving accuracy.
Solution Approach 2:
The patent adds structural dimensions to the parsing process by introducing parse trees and semantic frames as intermediate representations. These additional dimensional layers organize the temporal expression data hierarchically, enabling the system to maintain processing efficiency through structured data flow while accurately capturing complex temporal relationships that simple linear parsing would miss.
3Reliability
If basic grammar templates are used, then common temporal expressions are parsed correctly, but expressions with missing or ambiguous time components fail
Solution Approach 1:
The patent creates universal parse tree structures that can accommodate multiple types of temporal expressions through a single unified framework. The semantic frame extraction process is designed to handle various temporal expression formats (contiguous, non-contiguous, complete, incomplete) using the same underlying mechanism, making the system both reliable for standard expressions and adaptable to edge cases.
Solution Approach 2:
The patent dynamically adjusts parsing parameters based on the input expression characteristics. The system modifies its interpretation strategy depending on whether temporal components are present or missing, allowing it to maintain consistent reliability for complete expressions while adapting to handle ambiguous or incomplete temporal expressions through parameter adjustments in the semantic frame extraction process.
Data Source
AI summary
Embodiments provide for a temporal expression parser in a conversational data-to-text system are described herein. An example method may include receiving user query data comprising an input text string; generating, based at least in part on the input text string, a n-gram set comprising a plurality of n-gram elements; traversing each n-gram element in the n-gram set to generate a parse tree list comprising one or more parse trees based on a grammar template associated with the input text string; and generating, based at least in part on a last parse tree of the parse tree list, one or more semantic frames indicating a temporal expression associated with the input text string.