Extended Discourse Trees for Document Navigation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing keyword-based solutions fail to capture the relevance of different parts of text, leading to erroneous results in autonomous agents' responses to user queries, as they do not account for dialogue and rhetorical relationships within and between documents.

Innovation Solution

The creation of extended discourse trees that represent rhetorical relationships between discourse units within and across documents, using trained machine-learning models to classify entities and determine rhetorical relationships, allowing for improved navigation and search within bodies of text.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If keyword-based search solutions are used to determine relevance in formal documents, then the implementation is simple and fast, but the accuracy and coherence of results deteriorate because dialogue and rhetorical relationships are not captured

Engineering Contradiction:

Improveaccuracy of search resultsVSAvoidcomplexity of analysis technique

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments text into elementary discourse units (EDUs) and builds hierarchical discourse trees that capture rhetorical relationships between units. This segmentation approach transforms simple keyword matching into a structured analysis that preserves dialogue and logical relationships, thereby improving accuracy without requiring overly complex full-context analysis.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces discourse trees as an intermediary structure between keyword search and full semantic understanding. These trees serve as a mediator that captures rhetorical relationships and dialogue structure, enabling improved accuracy while maintaining computational feasibility through hierarchical organization rather than requiring complete contextual analysis.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If discourse trees are created for multiple documents and linked via rhetorical relationships to form extended discourse trees, then the coherence and relevance of text navigation improves, but the computational complexity and processing time increases

Engineering Contradiction:

Improvecoherence of discourse flowVSAvoidprocessing time for creating extended discourse trees

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent divides the large task of analyzing entire documents into smaller units by segmenting text into elementary discourse units and building individual discourse trees for each document. This segmentation allows parallel processing and reduces the computational burden compared to analyzing complete documents as single units, thereby reducing processing time while maintaining coherence.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent merges individual discourse trees into extended discourse trees by linking them through rhetorical relationships between matching elementary discourse units. This merging process preserves coherence across documents while building upon already-processed individual trees, avoiding redundant analysis and reducing overall processing time.

Inventive Principle:

Principle #5Merging (Combining)

3Measurement precision

If machine-learning models are used to classify entities and determine rhetorical relationships, then the precision of discourse analysis improves, but the computational resources and complexity required increases

Engineering Contradiction:

Improveprecision of entity classificationVSAvoidcomplexity of machine-learning implementation

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies machine-learning classification to segmented elementary discourse units rather than entire documents. This segmentation enables the use of simpler, more efficient classification models on small, focused units while achieving high overall precision through the cumulative effect of accurate unit-level classifications.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies different classification approaches at different levels of the discourse tree structure. Simple keyword matching may be used for basic entity identification while more sophisticated machine-learning models are applied only where needed for complex rhetorical relationship determination, optimizing resource usage while maintaining precision.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS11797773B2Navigating electronic documents using domain discourse trees

Publication Date: 2023.10.24 ORACLE INT CORP

AI summary

Navigating text using an extended discourse tree. In an example, a method accesses an extended discourse tree that includes a first discourse tree for a first document and a second discourse tree for a second document. The method determines a first elementary discourse unit that is responsive to a query from a user device and a corresponding first position. The method further determines a set of navigation options including a first rhetorical relationship between the first elementary discourse unit and a second elementary discourse unit of the first discourse tree and a second rhetorical relationship between the first elementary discourse unit and a third elementary discourse unit of the second discourse tree. The method presents the rhetorical relationships to a user device. Responsive to receiving, from a user device, a selection of a rhetorical relationship, the method presents a corresponding elementary discourse unit to the user device.