brady0127/sift1m-fanns-uniform

SIFT1M Dataset to Evaluate (Filtered) Approximate Nearest Neighbor Search — Uniform Variant

This dataset is intended to benchmark Approximate Nearest Neighbor Search (ANNS) and Filtered Approximate Nearest Neighbor Search (FANNS) algorithms. It is based on the classic SIFT1M dataset (1,000,000 base vectors and 10,000 query vectors, each 128-dimensional). Because SIFT1M ships without any structured metadata, we augment it with synthetic attributes drawn from a uniform distribution: every base vector receives one categorical label and one continuous range_value. Query attributes for three different filter types are sampled from the same uniform distribution as the base attributes. The ground truth for the up to k=100 nearest neighbors was computed for unfiltered ANNS and for filtered ANNS with the three different filter types. Please note that if fewer than 100 database items match the filter, then the ground truth can contain fewer than k=100 entries. This variant pairs with sift1m-fanns-normal (identical design, Gaussian attributes) to study the effect of attribute distribution.

Filter Types

Filter	Predicate	Description
label	label == required_label	Exact match on the categorical attribute.
range	range_low <= range_value <= range_high	Containment in a fixed-width interval.
both	label AND range	Conjunction of the two predicates above.

Parameters

Parameter	Value	Meaning
N	1,000,000	Number of base (database) vectors
nq	10,000	Number of query vectors
vec_dim	128	Embedding dimension (from SIFT1M)
NUM_LABELS	12	label ~ Uniform{0, 1, ..., 11}
range_value	Uniform[0, 1]	Continuous base attribute
RANGE_WIDTH	0.2	Width of every query range window (range_high - range_low)
GT_K	100	Max number of ground-truth neighbors per query
RANDOM_SEED	42	Fixed seed; the dataset is bit-reproducible

Expected Selectivity

Under the uniform distribution, every query has nearly the same selectivity (homogeneous), which makes this variant a clean control for isolating the effect of a single selectivity level.

Filter	Expected selectivity	Avg. matching items	Formula
label	8.33%	~83,333	1 / NUM_LABELS
range	20.00%	~200,000	RANGE_WIDTH
both	1.667%	~16,667	(1 / NUM_LABELS) × RANGE_WIDTH

Files and Description

File	Description
database_vectors.fvecs	128-dimensional base vectors. One vector per database item.
database_attributes.jsonl	JSON objects with {id, label, range_value} for each item. One JSON object per database item.
query_vectors.fvecs	128-dimensional query vectors. One vector per query.
ground_truth.ivecs	Ground truth for unfiltered nearest neighbor search. One vector per query.
label_query_attributes.jsonl	Query attributes for label (exact match) filtering. One JSON object per query.
ground_truth_label.ivecs	Ground truth for label-filtered NN search. One vector per query.
range_query_attributes.jsonl	Query attributes for range filtering. One JSON object per query.
ground_truth_range.ivecs	Ground truth for range-filtered NN search. One vector per query.
label_and_range_query_attributes.jsonl	Query attributes for the joint (label AND range) filter. One JSON object per query.
ground_truth_label_and_range.ivecs	Ground truth for joint-filtered NN search. One vector per query.

Formats

• .fvecs: Binary format for 32-bit floating point numbers (used for embedding vectors).
• .ivecs: Binary format for 32-bit signed integers (used for ground truth).
• .jsonl: Each line contains a JSON object (used for attributes).