Hierarchical Semantic Model for Document Chunk Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional search engines struggle to efficiently locate and highlight relevant information within documents, requiring users to open documents and perform additional searches, which is time-consuming and inefficient, especially when search keywords are not in the exact order or are scattered across different chunks.
Innovation Solution
A computer system generates a hierarchical semantic model of document content using metadata heuristics to identify and return specific 'chunks' that match search keywords, allowing for precise retrieval and display of relevant information without needing to open the entire document.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If conventional search engines return document names with one or two matching sentences, then the search process is simple, but the user cannot understand the meaning of search keywords in context and must open the document to find relevant information
Solution Approach 1:
The patent segments documents into meaningful chunks (paragraphs, sections, or semantically complete units) rather than treating documents as monolithic entities. This segmentation allows the system to return only the specific chunk containing relevant information, eliminating the need for users to open entire documents and manually search through them.
Solution Approach 2:
The patent introduces an intermediary processing layer between the search engine and the user that extracts and returns specific chunks from documents. This intermediary component analyzes document structure, identifies relevant chunks based on search keywords, and presents them directly to users, bridging the gap between simple search results and comprehensive document content.
2Measurement precision
If users open documents using native applications to search for keywords, then they can find exact matches, but the process is time-consuming and requires additional software operations
Solution Approach 1:
The patent performs preliminary indexing and chunking of documents before the user initiates a search. Document structures are pre-analyzed, and chunks are identified and stored in an optimized format. When a user searches, the system can quickly retrieve pre-processed chunks containing matching keywords without requiring users to open documents in native applications.
Solution Approach 2:
The patent replaces the mechanical process of opening documents in native applications with an automated electronic search system. Instead of requiring users to manually launch applications, navigate to files, and perform searches within those applications, the system automates the entire process by electronically scanning, analyzing, and returning relevant chunks directly through the search interface.
3Measurement precision
If search tools look for multiple keywords in exactly the same order, then the search is precise, but many relevant instances are missed when keywords appear in different orders
Solution Approach 1:
The patent implements dynamic keyword matching that adapts to different keyword arrangements within chunks. Rather than enforcing a fixed sequential order, the system dynamically identifies chunks containing all required keywords regardless of their position or order within the chunk, making the search both precise and flexible.
Solution Approach 2:
The patent changes the matching parameter from strict sequential keyword order to set-based keyword presence within a chunk. This parameter change allows the system to maintain precision by requiring all keywords to be present while gaining versatility by accepting any arrangement of those keywords within the identified chunk.
4Productivity
If users limit the search to a subset of keywords, then the search is faster, but many instances of the subset may exist and users spend significant effort finding the relevant content
Solution Approach 1:
The patent extracts and returns only the specific chunk containing the relevant subset of keywords, removing unnecessary surrounding content. This extraction approach maintains fast search execution by using efficient keyword matching while reducing the amount of content users must review, as only the relevant chunk is presented rather than entire documents or large portions of text.
Data Source
AI summary
A computer identifies a document in response to a search request from a user, wherein the document includes content data and metadata and the search request includes one or more search keywords. The computer generates a hierarchical semantic model of the content data of the document by applying heuristics to the metadata of the document. The computer identifies a chunk within the document by scanning the hierarchical semantic model, wherein the identified chunk includes a subset of the content data that satisfies the search keywords and the corresponding metadata. The computer returns the identified chunk to the requesting user.