Bi-Level Indexing for Graph Keyword Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing keyword search algorithms on graph-structured data face challenges such as inefficient use of indexing, lack of performance guarantees, and poor performance on certain graphs due to the difficulty in indexing connectivity and employing heuristic search strategies.
Innovation Solution
The implementation of a bi-level indexing scheme, BLINKS (Bi-Level INdexing for Keyword Search), which includes a cost-balanced expansion algorithm and a bi-level index structure that captures shortest-path information between nodes and keywords, allowing for bidirectional search and efficient pruning, and partitions the data graph into blocks to manage storage and search efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional keyword search algorithms are used on graph-structured data, then the search can be performed without complex indexing, but the query performance is poor and the number of accessed nodes is large
Solution Approach 1:
The patent segments the graph into hierarchical levels (L0, L1, L2, etc.) where L0 contains all nodes and subsequent levels contain progressively smaller subsets of nodes. This segmentation allows the search to prune entire levels once a threshold is exceeded, dramatically reducing the number of accessed nodes while maintaining manageable index complexity through the hierarchical structure.
Solution Approach 2:
The patent introduces a new dimensional structure by organizing graph nodes into hierarchical levels beyond the traditional single-level indexing. This multi-level hierarchy adds a vertical dimension to the index structure, enabling searches to eliminate entire levels efficiently and achieve better query performance without proportionally increasing indexing complexity.
2Reliability
If a naive index for graph connectivity is created, then connectivity information can be stored, but the storage requirement becomes unacceptably high (quadratic)
Solution Approach 1:
The patent segments the connectivity index into hierarchical levels where each level L contains only the necessary connectivity information for that level's nodes. This segmentation reduces storage from quadratic O(n²) to linear O(n) by storing connectivity information progressively across levels rather than maintaining complete connectivity for all nodes simultaneously.
Solution Approach 2:
The patent extracts only the essential connectivity information needed for each hierarchical level rather than storing complete graph connectivity. Each level L stores connectivity information only for nodes at that level and their relationships, removing redundant connectivity data that would otherwise require quadratic storage space.
3Ease of operation
If heuristic graph search strategies are employed, then the search can be conducted without strong performance guarantees, but the performance may be poor on certain graphs
Solution Approach 1:
The patent performs preliminary organization of graph nodes into hierarchical levels before conducting searches. This preliminary action creates a structured framework that guarantees performance by ensuring that searches can systematically prune levels based on predefined thresholds, rather than relying on heuristics that may fail on certain graph configurations.
Solution Approach 2:
The patent changes the search parameter from heuristic scoring to level-based threshold pruning. Instead of using heuristic strategies that estimate node importance, the system uses a deterministic parameter (level threshold) that guarantees performance by eliminating entire levels when the threshold is exceeded, providing reliable performance across all graph types.
Data Source
AI summary
Arrangements and methods for providing for the efficient implementation of ranked keyword searches on graph-structured data. Since it is difficult to directly build indexes for general schemaless graphs, conventional techniques highly rely on graph traversal in running time. The previous lack of more knowledge about graphs also resulted in great difficulties in applying pruning techniques. To address these problems, there is introduced herein a new scoring function while the block is used as an intermediate access level; the result is an opportunity to create sophisticated indexes for keyword search. Also proposed herein is a cost-balanced expansion algorithm to conduct a backward search, which provides a good theoretical guarantee in terms of the search cost.