Numeric Range Search Using Interval Index Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional search methods for numeric ranges in documents are inefficient due to the need for extensive comparisons across large datasets, and existing interval separation methods do not significantly improve search efficiency by not accounting for the number of numeric values within each interval.
Innovation Solution
A numeric range search device and method that utilize an interval index with records containing interval filters and reference information, allowing for efficient comparison by dividing the range into subintervals and generating interval queries to select records that match the input criteria, thereby reducing the number of comparisons needed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional keyword search or full-text search is used for numeric range search, then the search system is simple to implement, but it cannot comprehensively obtain documents conforming to numeric range criteria
Solution Approach 1:
The patent segments numeric values into exponential parts and mantissa parts, creating separate index structures for each. This segmentation allows the search system to handle numeric ranges by comparing exponential parts first, then mantissa parts, achieving reliable numeric range search without requiring complex full-text search modifications.
Solution Approach 2:
The patent adds a new dimension to the search system by creating a separate exponential part index alongside the traditional full-text index. This dimensional addition enables efficient numeric range search through exponential part comparison while maintaining compatibility with existing full-text search capabilities.
2Productivity
If numeric values are indexed by dividing into exponential part and mantissa part, then numeric range search efficiency is improved, but the index structure becomes more complex
Solution Approach 1:
The patent segments numeric values into exponential parts and mantissa parts, storing them in separate index structures. This segmentation enables efficient numeric range search by first filtering using exponential parts, then refining results using mantissa parts, significantly improving search speed despite increased index complexity.
Solution Approach 2:
The patent performs preliminary action by pre-computing and storing exponential parts and mantissa parts in separate index structures during document indexing. This preliminary processing enables fast numeric range search execution, as the search operation only needs to query the pre-organized exponential part index rather than parsing and comparing full numeric values during search.
3Reliability
If all numeric ranges in large datasets are compared against search criteria, then search completeness is ensured, but search time increases significantly
Solution Approach 1:
The patent segments the comparison process into two stages: first comparing exponential parts to quickly filter out non-matching ranges, then comparing mantissa parts only for candidates that pass the exponential part filter. This segmented approach ensures search completeness while dramatically reducing search time by avoiding full comparisons for all numeric ranges in large datasets.
Solution Approach 2:
The patent applies partial action by performing complete numeric range comparison only on a subset of candidates that first match the exponential part criteria. This partial comparison approach maintains search completeness for the final result while reducing overall search time by avoiding unnecessary full comparisons for clearly non-matching entries.
Data Source
AI summary
A numeric range search device includes: interval index storage means for storing, for numeric ranges to be searched, an interval index that is made up of records each including at least an interval filter and reference information and in which the records are organized in units of records whose interval filters are at least partially common, the interval filter being data indicating a correspondence relationship between an input numeric value or numeric range and subintervals obtained by dividing a range including all of the numeric ranges by a predetermined boundary value, and the reference information being information for referring to a numeric range to be searched; interval query generation means for generating an interval query which is data indicating a correspondence relationship between the input numeric value or numeric range and the subintervals; and candidate selection means for selecting, from the records, a record for which an AND of an interval filter included in the record and the interval query is equal to the interval query.