Database Search Speed and Quality via Metadata Caching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Database searching techniques face challenges in quickly and accurately querying both standard and user-provided metadata without incurring computationally intensive post-index filtering actions or returning out-of-date results, due to the computational expense of indexing large collections and real-time changes in user-specific metadata.
Innovation Solution
Mapping fields from multiple tables of different types to a single table, indexing these mapped fields for a single index, and executing queries using this index to output results, while also executing a separate query for recent search activity to combine results and account for user-provided metadata changes since the last index update.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If indexing is performed on large collections of documents to enable fast search, then search speed is improved, but computational cost and complexity increase significantly
Solution Approach 1:
The patent segments the indexing process by creating separate indexes for different types of metadata fields. Standard metadata fields (title, author, subject) are indexed separately from user-provided metadata fields (tags, ratings, comments). This segmentation allows the system to optimize indexing strategies for each type independently, reducing overall complexity while maintaining fast search performance.
Solution Approach 2:
The patent applies partial indexing by not indexing all possible fields equally. Instead, it selectively indexes standard metadata fields that are frequently queried, while user-provided metadata fields are handled through a separate mechanism. This partial action approach reduces the computational burden of full indexing while still providing fast search capabilities for the most important fields.
2Measurement precision
If standard metadata is indexed for fast querying, then search accuracy is improved, but user-specific metadata queries become slower
Solution Approach 1:
The patent merges the querying process for standard metadata and user-provided metadata into a unified search operation. Instead of requiring separate indexing for user-specific fields, the system combines results from the standard metadata index with user-provided metadata from a separate storage structure, allowing both types of queries to be handled efficiently in a single operation.
Solution Approach 2:
The patent introduces an intermediary layer that handles user-provided metadata separately from the main index. This intermediary structure (such as a hash map or separate data structure) stores user-provided metadata and enables fast retrieval without requiring full re-indexing of the document collection. The intermediary acts as a mediator between the main index and user-specific queries.
3Reliability
If user-provided metadata is indexed in real-time, then query freshness is improved, but computational overhead increases
Solution Approach 1:
The patent applies preliminary action by pre-processing and storing user-provided metadata in an optimized structure as it is generated, rather than waiting for batch indexing operations. When user metadata is added or modified, it is immediately stored in the intermediary structure, making it available for queries without requiring subsequent re-indexing. This preliminary action ensures query freshness while minimizing computational overhead.
Data Source
AI summary
An apparatus may map one or more fields from a plurality of tables having a first table type to one or more mapped fields of a single table having a second table type and index the one or more mapped fields to correspond to a single index for the single table having the second table type. The apparatus may execute a first query using the single index corresponding to the one or more mapped fields to output a result for the first query. The apparatus may execute a second query for search activity that has occurred on the database since a last update to the single index and combine the result for the first query with the search activity that has occurred on the database since the last update to the single index to output a combined result based on the second query.