Dataset guide 05
Bitcoin Bloom Filter Dataset
A Bloom filter is a compact probabilistic data structure that can quickly answer whether a value is definitely absent or possibly present in a dataset.
What a Bloom-filter result means
A negative lookup is definitive for the dataset version used to construct the filter. A positive lookup may be a false positive and must be checked against the accompanying exact address, hash or UTXO dataset.
The false-positive probability depends on the number of inserted records, filter size and number of hash functions. Those parameters must be published with every release.
Where filters are useful
Bloom filters are useful when an application must test large numbers of candidate values without holding a much larger exact database in memory. They can reduce unnecessary disk lookups in research pipelines, monitoring tools and data-processing systems.
Versioning and reproducibility
A filter is only valid for the source snapshot from which it was generated. The manifest should bind it to a Bitcoin block height, input checksum, element count, bit size, hash count and serialization version.
Frequently asked
Questions about bitcoin bloom filter dataset
Can a Bloom filter return a wrong negative?
A correctly constructed Bloom filter does not produce false negatives for the values inserted into that specific version.
Can it replace the exact dataset?
No. Positive results require exact verification, and the filter does not contain balances or other record fields.