Document Portion Layout Prediction for Targeted Search

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current document search technologies are inefficient when searching for keywords in documents without layout information, as they lack the ability to accurately predict and utilize document layout types for targeted searches.

Innovation Solution

A machine learning system predicts the layout type of document portions and indexes them, allowing for efficient searching by layout through layout-specific collections of index entries, even in documents without explicit layout tags.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If document search is performed without layout information, then search coverage is comprehensive, but search efficiency deteriorates due to inability to target specific layout types

Engineering Contradiction:

Improvesearch efficiencyVSAvoidlayout information

Core Design Contradiction:

ProductivityVSLoss of information

Solution Approach 1:

The system performs preliminary layout prediction and indexing before the actual search operation. By predicting layout types and creating layout-specific index collections in advance, the system prepares the search infrastructure proactively, allowing efficient layout-targeted searches to be executed without losing layout information during the search process itself

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The index is segmented into multiple layout-specific collections (e.g., title collection, body text collection, caption collection, list collection, table collection, footnote collection) based on predicted layout types. This segmentation allows the search system to target specific layout collections during querying, improving search efficiency while maintaining comprehensive coverage across all document portions

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If layout prediction is performed for all document portions, then search precision is improved, but processing time increases

Engineering Contradiction:

Improvelayout prediction precisionVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system applies layout prediction selectively to document portions that are likely to contain search-relevant content, rather than uniformly processing every single portion. The machine learning model predicts layout types for portions where layout information would be most valuable for search operations, balancing precision improvement with processing time constraints

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS12072935B2Machine-learning of document portion layout

Publication Date: 2024.08.27 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

Machine learning to predict a layout type that each of a plurality of portions of a document appears in. This is done even though the computer-readable representation of the document does not contain information at the granularity of the prediction to be made that identifies which layout type that each of the plurality of document portions belongs in. For each of a plurality of the portions, the machine-learning system predicts the layout type that the respective portion appears in, and indexes the document using the predictions so as to result in a computer-readable index. The index represents a predicted layout type associated with each of the plurality of portions of the document. Thus, the index can be used to search based on position of a searched term within the document.