Efficient Direct-Access Ranked Retrieval

This repository contains implementations and experimental code for efficient direct-access ranked retrieval algorithms. For detailed technical information, see the Technical Report.

Repository Structure

Core Implementations

methods/kth/

Contains implementations of k-th element retrieval algorithms:

• EpsRange - Epsilon-based range approach
• EpsHier - Epsilon-based hierarchical method
• KthLevel - Level-based k-th element algorithm
• Baselines: TA (Threshold Algorithm) and Fagin

methods/range_search/

Contains implementations of range search algorithms:

• Hierarchical Sampling - Novel hierarchical sampling approach
• Baseline algorithms: Partition Tree, KD-tree, R-tree

methods/utils.py

Utility functions and helper methods used across different algorithms.

Experimental Code

experiments/

Example codes and benchmarks demonstrating algorithm usage:

• Performance comparisons and analysis
• Real-world dataset evaluations
• Memory usage studies
• Visualization tools

Supporting Files

ranges/

Range-related data structures and utilities:

• Base range implementations
• Stripe range operations