Integrating semantic and content features into bloom filters for efficient image retrieval.

Area of Science:

• Computer Science
• Data Structures
• Information Retrieval

Background:

• Bloom filters (BFs) are probabilistic data structures for efficient set membership queries.
• BFs can exhibit high false positive rates, especially with large datasets, limiting their practical use.
• Existing methods often struggle to balance accuracy with memory efficiency in large-scale applications.

Purpose of the Study:

• To develop a novel method for reducing false positives in Bloom filters for image retrieval.
• To enhance the compactness and performance of Bloom filters by incorporating auxiliary data.
• To improve the efficiency of Bloom filters for large-scale data management and retrieval tasks.

Main Methods:

• Incorporated content-based check bits derived from image captions and image intensity into Bloom filters.
• Generated check bits from both textual descriptions and visual data of images.
• Evaluated the method on a dataset of 31,783 images from Flickr.

Main Results:

• Significantly reduced the false positive rate in Bloom filters.
• Achieved higher efficiency in terms of memory usage and processing power.
• Demonstrated effectiveness even with smaller Bloom filter sizes.
• Validated the approach on a substantial image dataset.

Conclusions:

• The proposed method effectively reduces false positives in Bloom filters for image retrieval.
• Integrating auxiliary data sources like image captions and intensity enhances Bloom filter performance and compactness.
• This technique offers a practical solution for large-scale image retrieval, database management, and network security, minimizing computational overhead.