Contextual Summary Information Across Documents

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing information retrieval systems face limitations in computing summary information for complex document structures, often resulting in low factual relationships between user queries and document properties, leading to noisy and inaccurate summary information.

Innovation Solution

A method is introduced to compute contextual summary information by selecting query-dependent subsections of documents, associating relevant document properties, and analyzing these properties to improve the accuracy and relevance of summary information, enabling better navigation, question-answering, and query disambiguation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If summary information is computed across entire documents without contextual selection, then the coverage of document content is comprehensive, but the factual relationship between query and document properties deteriorates, resulting in noisy and inaccurate summary information

Engineering Contradiction:

Improveaccuracy of summary informationVSAvoidcontextual relevance

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The patent divides documents into relevant and irrelevant segments based on query matching. Only document portions containing query matches or contextually related information are selected for summary computation, eliminating noise from unrelated document sections while preserving factual relationships between queries and document properties.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies different processing quality to different parts of documents. High-quality contextual analysis and property extraction are applied only to query-relevant document sections, while irrelevant sections are excluded. This local quality approach improves overall summary accuracy by focusing computational resources on factually relevant content.

Inventive Principle:

Principle #3Local quality

2Productivity

If a priori technology is used for summary information computation, then the processing is simple and fast, but the capability to handle complex document structures deteriorates, leading to substantial limitations in quality

Engineering Contradiction:

Improveprocessing efficiencyVSAvoidhandling of complex document structures

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent performs preliminary actions by first identifying query matches in documents, then selecting document portions based on these matches before computing summary information. This preliminary contextual selection enables efficient processing of complex document structures by pre-filtering relevant content, maintaining productivity while improving adaptability to various document formats and complexities.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces dynamic document portion selection based on query context. Instead of static a priori processing, the system dynamically determines which document sections are relevant to the specific query, allowing flexible adaptation to complex document structures while maintaining processing efficiency through targeted analysis of only necessary portions.

Inventive Principle:

Principle #15Dynamics

3Measurement precision

If document properties are selected without query-dependent subsection selection, then the extraction process is straightforward, but the contextual relationship between query and document properties deteriorates, resulting in low factual relevance

Engineering Contradiction:

Improvefactual relationship qualityVSAvoidprocessing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts only the necessary document properties from query-relevant portions rather than processing entire documents. By taking out and analyzing only the properties associated with matched document sections, the system improves factual relationship quality while managing processing complexity through selective extraction rather than comprehensive analysis.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies partial action by computing summary information only for document portions that contain query matches or contextual relationships, rather than analyzing entire documents. This partial processing approach improves the factual relevance of extracted properties while keeping processing complexity manageable by focusing computational effort on relevant sections only.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS7912849B2Method for determining contextual summary information across documents

Publication Date: 2011.03.22 MICROSOFT CORP

AI summary

In a method for determining contextual summary information across documents retrieved in response to a user query applied to a collection of documents the documents matching the query are identified. A query-dependent subsection of each of the matching documents is selected. Document properties associated with the document subsection are selected and associated with localized structures within the document. Relationships between localized document properties and user queries are determined and used to compute contextual summary information, whereby localized document properties are profiled across the retrieved documents in a contextual manner. The method allows a user query to select localized structures within a matching document and is generally applicable in information retrieval and the analysis of retrieved information.