Index Tree Spatial Search for Business Delivery Areas
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current spatial search methods for businesses, such as those using PostgreSQL and Lucene, become inefficient when dealing with large numbers of delivery areas, leading to errors and poor performance as spatial data increases.
Innovation Solution
A spatial location search solution based on an index tree and circumscribed rectangle filtering, where two index trees are constructed using lower left and upper right vertexes of delivery areas, allowing for efficient aggregation and identification of nearby businesses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If spatial query is carried out by comparing delivery areas of all businesses one by one with user location, then the solution is simple to implement, but the search efficiency deteriorates when the number of businesses reaches tens of thousands
Solution Approach 1:
The patent segments the continuous delivery area into multiple discrete grids. Each grid can be independently indexed and queried, transforming a complex continuous spatial comparison into simpler discrete grid operations. This segmentation enables efficient filtering of candidate businesses by only comparing user location against relevant grids rather than all delivery areas.
Solution Approach 2:
The patent introduces a grid dimension overlay on top of the geographic coordinate system. By mapping delivery areas and user locations onto this grid dimension, the system creates an additional indexing layer that accelerates spatial queries. The grid dimension acts as a preprocessing structure that reduces the search space before detailed geometric comparisons are performed.
2Productivity
If delivery areas are divided into grids for spatial query using Lucene, then the search performance is improved, but the amount of spatial data increases to one million or ten million causing index errors
Solution Approach 1:
The patent extracts only the essential geometric characteristics of delivery areas by representing them as circumscribed rectangles defined by minimum and maximum latitude and longitude values. This extraction eliminates the need to store and index entire polygon geometries or detailed boundary information, retaining only the critical bounding box data needed for spatial filtering.
Solution Approach 2:
The patent transforms the complex spatial representation of delivery areas into simplified parameter sets (min_lat, max_lat, min_lng, max_lng). This parameter transformation converts continuous geometric data into discrete numeric ranges that are more compact and suitable for efficient indexing and comparison operations in the grid-based system.
3Reliability
If the index is constructed with one million or ten million spatial data points, then comprehensive coverage is achieved, but the index size becomes too large causing query errors
Solution Approach 1:
The patent segments the large-scale spatial index into manageable grid units. Instead of constructing a single massive index with all spatial data points, the system divides the search space into grids and maintains separate, smaller indexes for each grid or grid cluster. This segmentation reduces memory requirements and avoids index size limitations while preserving comprehensive coverage through the collective set of grid indexes.
Data Source
AI summary
A method of searching for a business by constructing an index tree using a geographic location of a user is provided. In some examples, the method includes obtaining the geographic location of the user, and retrieving circumscribed rectangles covering the geographic location with a pre-constructed index tree, where the circumscribed rectangles are a smallest circumscribed rectangles covering delivery areas of businesses (101). Target delivery areas covering the geographic location are determined from the delivery areas corresponding to the circumscribed rectangles(102). Businesses corresponding to the target delivery areas are determined as deliverable businesses corresponding to the geographic location(103). The pre-constructed index tree includes a first index tree constructed based on location data of lower left vertexes of circumscribed rectangles of delivery areas of the businesses, and a second index tree constructed based on location data of upper right vertexes of the circumscribed rectangles of the delivery areas of the business.