Search Query Refinement via Archetype-Based Chunking
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 with a graphical user interface that selects an archetype for search keywords, identifies relevant chunks within documents, and returns these chunks to the user, utilizing a query engine, cache engine, and stream engine to process and retrieve documents from various data sources, applying semantic and contextual analysis to highlight matching terms.
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 and fast, but the user cannot understand the meaning of search keywords in context and must open documents to find relevant information
Solution Approach 1:
The patent segments documents into meaningful chunks (paragraphs, sections, or semantic units) and identifies relevant chunks that contain search keywords. Instead of returning entire documents or isolated sentences, the system returns specific chunks that are most relevant to the search query, making it easier for users to find information without opening full documents.
Solution Approach 2:
The patent introduces an intermediary processing layer between the search engine and the user. This layer analyzes document content, identifies relevant chunks, and presents them in a structured format that provides context while saving time. The intermediary process includes semantic analysis, keyword matching, and chunk selection to bridge the gap between simple search results and comprehensive document review.
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 the native application does not highlight relevant portions
Solution Approach 1:
The patent performs preliminary actions by pre-processing documents into chunks and pre-identifying relevant portions before the user searches. The system analyzes document structure, segments content meaningfully, and prepares index structures that enable fast retrieval of relevant chunks. This preliminary processing eliminates the need for users to manually search through entire documents in native applications.
Solution Approach 2:
The patent uses visual highlighting (analogous to color changes) to emphasize relevant chunks and matching keywords within the returned results. Relevant portions are visually distinguished through formatting, highlighting, or emphasis markers that make it immediately obvious which parts of the document contain the search keywords, eliminating the need for users to manually scan or highlight text.
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 introduces dynamic keyword matching that adapts to different keyword arrangements within chunks. Instead of requiring fixed sequential order, the system dynamically identifies chunks containing all search keywords regardless of their order, while maintaining precision by ensuring all keywords are present within the same relevant chunk. The matching process is flexible yet precise, adapting to various keyword configurations.
Solution Approach 2:
The patent changes the matching parameter from strict sequential order to set-based inclusion. Instead of requiring keywords to appear in the exact order specified by the user, the system checks whether all keywords are present within a chunk as a set, regardless of their sequence. This parameter change maintains precision (all keywords must be present) while increasing adaptability to different keyword arrangements.
4Ease of operation
If users limit the search to a subset of keywords, then the search is faster and easier, but many relevant instances are scattered across documents and require significant effort to find
Solution Approach 1:
The patent implements feedback mechanisms that analyze search results and provide information about the relevance and distribution of matching chunks. The system returns structured results that indicate how many chunks were found, their relevance scores, and their locations within documents. This feedback helps users assess whether to refine their search or review the returned chunks, reducing the time needed to find relevant information among multiple instances.
Data Source
AI summary
After receiving a search keyword provided by a user, a computer selects an archetype for the search keyword. The computer identifies one or more search results in accordance with the archetype and returns at least one of the search results to the user. After selecting the archetype, the computer identifies at least one query operator for the selected archetype, constructs a search query using the query operator, and executes the search query against one or more data sources. Sometimes, the computer solicits user instructions with respect to the archetype and then generates feedback to the user instructions. This process may repeat multiple loops until the user submits a search query execution request, which suggests that the user is satisfied with the customized search query.