Add binary metric types SUBSTRUCTURE/SUPERSTRUCTURE back (#26766)

Signed-off-by: Yudong Cai <yudong.cai@zilliz.com>

internal/core/unittest/test_similarity_corelation.cpp

@@ -20,4 +20,6 @@ TEST(SimilarityCorelation, Naive) {
     ASSERT_FALSE(milvus::PositivelyRelated(knowhere::metric::L2));
     ASSERT_FALSE(milvus::PositivelyRelated(knowhere::metric::HAMMING));
     ASSERT_FALSE(milvus::PositivelyRelated(knowhere::metric::JACCARD));
+    ASSERT_FALSE(milvus::PositivelyRelated(knowhere::metric::SUBSTRUCTURE));
+    ASSERT_FALSE(milvus::PositivelyRelated(knowhere::metric::SUPERSTRUCTURE));
 }

internal/proxy/util.go

@@ -446,7 +446,7 @@ func validateMetricType(dataType schemapb.DataType, metricTypeStrRaw string) err
 		if dataType == schemapb.DataType_FloatVector {
 			return nil
 		}
-	case metric.JACCARD, metric.HAMMING:
+	case metric.JACCARD, metric.HAMMING, metric.SUBSTRUCTURE, metric.SUPERSTRUCTURE:
 		if dataType == schemapb.DataType_BinaryVector {
 			return nil
 		}

pkg/util/indexparamcheck/bin_flat_checker_test.go

@@ -31,6 +31,7 @@ func Test_binFlatChecker_CheckTrain(t *testing.T) {
 		DIM:    strconv.Itoa(128),
 		Metric: metric.COSINE,
 	}
+
 	p4 := map[string]string{
 		DIM:    strconv.Itoa(128),
 		Metric: metric.HAMMING,
@@ -39,6 +40,14 @@ func Test_binFlatChecker_CheckTrain(t *testing.T) {
 		DIM:    strconv.Itoa(128),
 		Metric: metric.JACCARD,
 	}
+	p6 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		Metric: metric.SUBSTRUCTURE,
+	}
+	p7 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		Metric: metric.SUPERSTRUCTURE,
+	}
 
 	cases := []struct {
 		params   map[string]string
@@ -51,6 +60,8 @@ func Test_binFlatChecker_CheckTrain(t *testing.T) {
 		{p3, false},
 		{p4, true},
 		{p5, true},
+		{p6, true},
+		{p7, true},
 	}
 
 	c := newBinFlatChecker()

pkg/util/indexparamcheck/bin_ivf_flat_checker.go

@@ -9,8 +9,8 @@ type binIVFFlatChecker struct {
 }
 
 func (c binIVFFlatChecker) StaticCheck(params map[string]string) error {
-	if !CheckStrByValues(params, Metric, BinMetrics) {
+	if !CheckStrByValues(params, Metric, BinIvfMetrics) {
-		return fmt.Errorf("metric type not found or not supported, supported: %v", BinMetrics)
+		return fmt.Errorf("metric type not found or not supported, supported: %v", BinIvfMetrics)
 	}
 
 	if !CheckIntByRange(params, NLIST, MinNList, MaxNList) {

pkg/util/indexparamcheck/bin_ivf_flat_checker_test.go

@@ -65,6 +65,7 @@ func Test_binIVFFlatChecker_CheckTrain(t *testing.T) {
 		IVFM:   strconv.Itoa(4),
 		NBITS:  strconv.Itoa(8),
 	}
+
 	p4 := map[string]string{
 		DIM:    strconv.Itoa(128),
 		Metric: metric.HAMMING,
@@ -79,6 +80,20 @@ func Test_binIVFFlatChecker_CheckTrain(t *testing.T) {
 		IVFM:   strconv.Itoa(4),
 		NBITS:  strconv.Itoa(8),
 	}
+	p6 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		Metric: metric.SUBSTRUCTURE,
+		NLIST:  strconv.Itoa(100),
+		IVFM:   strconv.Itoa(4),
+		NBITS:  strconv.Itoa(8),
+	}
+	p7 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		Metric: metric.SUPERSTRUCTURE,
+		NLIST:  strconv.Itoa(100),
+		IVFM:   strconv.Itoa(4),
+		NBITS:  strconv.Itoa(8),
+	}
 
 	cases := []struct {
 		params   map[string]string
@@ -96,6 +111,8 @@ func Test_binIVFFlatChecker_CheckTrain(t *testing.T) {
 
 		{p4, true},
 		{p5, true},
+		{p6, false},
+		{p7, false},
 	}
 
 	c := newBinIVFFlatChecker()

pkg/util/indexparamcheck/binary_vector_base_checker.go

@@ -13,8 +13,8 @@ type binaryVectorBaseChecker struct {
 }
 
 func (c binaryVectorBaseChecker) staticCheck(params map[string]string) error {
-	if !CheckStrByValues(params, Metric, BinMetrics) {
+	if !CheckStrByValues(params, Metric, BinIDMapMetrics) {
-		return fmt.Errorf("metric type not found or not supported, supported: %v", BinMetrics)
+		return fmt.Errorf("metric type not found or not supported, supported: %v", BinIDMapMetrics)
 	}
 
 	return nil

pkg/util/indexparamcheck/constraints.go

@@ -42,10 +42,11 @@ const (
 var METRICS = []string{metric.L2, metric.IP, metric.COSINE} // const
 
 // BinIDMapMetrics is a set of all metric types supported for binary vector.
-var BinMetrics = []string{metric.HAMMING, metric.JACCARD}                                       // const
+var BinIDMapMetrics = []string{metric.HAMMING, metric.JACCARD, metric.SUBSTRUCTURE, metric.SUPERSTRUCTURE} // const
-var HnswMetrics = []string{metric.L2, metric.IP, metric.COSINE, metric.HAMMING, metric.JACCARD} // const
+var BinIvfMetrics = []string{metric.HAMMING, metric.JACCARD}                                               // const
-var supportDimPerSubQuantizer = []int{32, 28, 24, 20, 16, 12, 10, 8, 6, 4, 3, 2, 1}             // const
+var HnswMetrics = []string{metric.L2, metric.IP, metric.COSINE, metric.HAMMING, metric.JACCARD}            // const
-var supportSubQuantizer = []int{96, 64, 56, 48, 40, 32, 28, 24, 20, 16, 12, 8, 4, 3, 2, 1}      // const
+var supportDimPerSubQuantizer = []int{32, 28, 24, 20, 16, 12, 10, 8, 6, 4, 3, 2, 1}                        // const
+var supportSubQuantizer = []int{96, 64, 56, 48, 40, 32, 28, 24, 20, 16, 12, 8, 4, 3, 2, 1}                 // const
 
 const (
 	FloatVectorDefaultMetricType  = metric.IP

pkg/util/indexparamcheck/diskann_checker_test.go

@@ -37,6 +37,7 @@ func Test_diskannChecker_CheckTrain(t *testing.T) {
 		DIM:    strconv.Itoa(128),
 		Metric: metric.COSINE,
 	}
+
 	p4 := map[string]string{
 		DIM:    strconv.Itoa(128),
 		Metric: metric.HAMMING,
@@ -45,6 +46,14 @@ func Test_diskannChecker_CheckTrain(t *testing.T) {
 		DIM:    strconv.Itoa(128),
 		Metric: metric.JACCARD,
 	}
+	p6 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		Metric: metric.SUBSTRUCTURE,
+	}
+	p7 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		Metric: metric.SUPERSTRUCTURE,
+	}
 
 	cases := []struct {
 		params   map[string]string
@@ -59,6 +68,8 @@ func Test_diskannChecker_CheckTrain(t *testing.T) {
 		{p3, true},
 		{p4, false},
 		{p5, false},
+		{p6, false},
+		{p7, false},
 	}
 
 	c := newDiskannChecker()

pkg/util/indexparamcheck/flat_checker_test.go

@@ -23,6 +23,7 @@ func Test_flatChecker_CheckTrain(t *testing.T) {
 		DIM:    strconv.Itoa(128),
 		Metric: metric.COSINE,
 	}
+
 	p4 := map[string]string{
 		DIM:    strconv.Itoa(128),
 		Metric: metric.HAMMING,
@@ -31,7 +32,14 @@ func Test_flatChecker_CheckTrain(t *testing.T) {
 		DIM:    strconv.Itoa(128),
 		Metric: metric.JACCARD,
 	}
-
+	p6 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		Metric: metric.SUBSTRUCTURE,
+	}
+	p7 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		Metric: metric.SUPERSTRUCTURE,
+	}
 	cases := []struct {
 		params   map[string]string
 		errIsNil bool
@@ -41,6 +49,8 @@ func Test_flatChecker_CheckTrain(t *testing.T) {
 		{p3, true},
 		{p4, false},
 		{p5, false},
+		{p6, false},
+		{p7, false},
 	}
 
 	c := newFlatChecker()

pkg/util/indexparamcheck/hnsw_checker_test.go

@@ -49,6 +49,7 @@ func Test_hnswChecker_CheckTrain(t *testing.T) {
 		EFConstruction: strconv.Itoa(200),
 		Metric:         metric.COSINE,
 	}
+
 	p4 := map[string]string{
 		DIM:            strconv.Itoa(128),
 		HNSWM:          strconv.Itoa(16),
@@ -61,6 +62,18 @@ func Test_hnswChecker_CheckTrain(t *testing.T) {
 		EFConstruction: strconv.Itoa(200),
 		Metric:         metric.JACCARD,
 	}
+	p6 := map[string]string{
+		DIM:            strconv.Itoa(128),
+		HNSWM:          strconv.Itoa(16),
+		EFConstruction: strconv.Itoa(200),
+		Metric:         metric.SUBSTRUCTURE,
+	}
+	p7 := map[string]string{
+		DIM:            strconv.Itoa(128),
+		HNSWM:          strconv.Itoa(16),
+		EFConstruction: strconv.Itoa(200),
+		Metric:         metric.SUPERSTRUCTURE,
+	}
 
 	cases := []struct {
 		params   map[string]string
@@ -76,6 +89,8 @@ func Test_hnswChecker_CheckTrain(t *testing.T) {
 		{p3, true},
 		{p4, true},
 		{p5, true},
+		{p6, false},
+		{p7, false},
 	}
 
 	c := newHnswChecker()

pkg/util/indexparamcheck/ivf_base_checker_test.go

@@ -32,6 +32,7 @@ func Test_ivfBaseChecker_CheckTrain(t *testing.T) {
 		NLIST:  strconv.Itoa(1024),
 		Metric: metric.COSINE,
 	}
+
 	p4 := map[string]string{
 		DIM:    strconv.Itoa(128),
 		NLIST:  strconv.Itoa(1024),
@@ -42,6 +43,16 @@ func Test_ivfBaseChecker_CheckTrain(t *testing.T) {
 		NLIST:  strconv.Itoa(1024),
 		Metric: metric.JACCARD,
 	}
+	p6 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		NLIST:  strconv.Itoa(1024),
+		Metric: metric.SUBSTRUCTURE,
+	}
+	p7 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		NLIST:  strconv.Itoa(1024),
+		Metric: metric.SUPERSTRUCTURE,
+	}
 
 	cases := []struct {
 		params   map[string]string
@@ -55,6 +66,8 @@ func Test_ivfBaseChecker_CheckTrain(t *testing.T) {
 		{p3, true},
 		{p4, false},
 		{p5, false},
+		{p6, false},
+		{p7, false},
 	}
 
 	c := newIVFBaseChecker()

pkg/util/indexparamcheck/ivf_pq_checker_test.go

@@ -84,6 +84,7 @@ func Test_ivfPQChecker_CheckTrain(t *testing.T) {
 		NBITS:  strconv.Itoa(8),
 		Metric: metric.COSINE,
 	}
+
 	p4 := map[string]string{
 		DIM:    strconv.Itoa(128),
 		NLIST:  strconv.Itoa(1024),
@@ -98,6 +99,20 @@ func Test_ivfPQChecker_CheckTrain(t *testing.T) {
 		NBITS:  strconv.Itoa(8),
 		Metric: metric.JACCARD,
 	}
+	p6 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		NLIST:  strconv.Itoa(1024),
+		IVFM:   strconv.Itoa(4),
+		NBITS:  strconv.Itoa(8),
+		Metric: metric.SUBSTRUCTURE,
+	}
+	p7 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		NLIST:  strconv.Itoa(1024),
+		IVFM:   strconv.Itoa(4),
+		NBITS:  strconv.Itoa(8),
+		Metric: metric.SUPERSTRUCTURE,
+	}
 
 	cases := []struct {
 		params   map[string]string
@@ -120,6 +135,8 @@ func Test_ivfPQChecker_CheckTrain(t *testing.T) {
 		{p3, true},
 		{p4, false},
 		{p5, false},
+		{p6, false},
+		{p7, false},
 	}
 
 	c := newIVFPQChecker()

pkg/util/indexparamcheck/ivf_sq_checker_test.go

@@ -46,6 +46,7 @@ func Test_ivfSQChecker_CheckTrain(t *testing.T) {
 		NBITS:  strconv.Itoa(8),
 		Metric: metric.COSINE,
 	}
+
 	p4 := map[string]string{
 		DIM:    strconv.Itoa(128),
 		NLIST:  strconv.Itoa(100),
@@ -58,6 +59,18 @@ func Test_ivfSQChecker_CheckTrain(t *testing.T) {
 		NBITS:  strconv.Itoa(8),
 		Metric: metric.JACCARD,
 	}
+	p6 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		NLIST:  strconv.Itoa(100),
+		NBITS:  strconv.Itoa(8),
+		Metric: metric.SUBSTRUCTURE,
+	}
+	p7 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		NLIST:  strconv.Itoa(100),
+		NBITS:  strconv.Itoa(8),
+		Metric: metric.SUPERSTRUCTURE,
+	}
 
 	cases := []struct {
 		params   map[string]string
@@ -73,6 +86,8 @@ func Test_ivfSQChecker_CheckTrain(t *testing.T) {
 		{p3, true},
 		{p4, false},
 		{p5, false},
+		{p6, false},
+		{p7, false},
 	}
 
 	c := newIVFSQChecker()

pkg/util/indexparamcheck/raft_ivf_pq_checker_test.go

@@ -77,6 +77,7 @@ func Test_raftIVFPQChecker_CheckTrain(t *testing.T) {
 		NBITS:  strconv.Itoa(8),
 		Metric: metric.COSINE,
 	}
+
 	p4 := map[string]string{
 		DIM:    strconv.Itoa(128),
 		NLIST:  strconv.Itoa(1024),
@@ -91,6 +92,20 @@ func Test_raftIVFPQChecker_CheckTrain(t *testing.T) {
 		NBITS:  strconv.Itoa(8),
 		Metric: metric.JACCARD,
 	}
+	p6 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		NLIST:  strconv.Itoa(1024),
+		IVFM:   strconv.Itoa(4),
+		NBITS:  strconv.Itoa(8),
+		Metric: metric.SUBSTRUCTURE,
+	}
+	p7 := map[string]string{
+		DIM:    strconv.Itoa(128),
+		NLIST:  strconv.Itoa(1024),
+		IVFM:   strconv.Itoa(4),
+		NBITS:  strconv.Itoa(8),
+		Metric: metric.SUPERSTRUCTURE,
+	}
 
 	cases := []struct {
 		params   map[string]string
@@ -112,6 +127,8 @@ func Test_raftIVFPQChecker_CheckTrain(t *testing.T) {
 		{p3, true},
 		{p4, false},
 		{p5, false},
+		{p6, false},
+		{p7, false},
 	}
 
 	c := newRaftIVFPQChecker()

pkg/util/metric/metric_type.go

@@ -30,4 +30,10 @@ const (
 
 	// JACCARD represents jaccard distance
 	JACCARD MetricType = "JACCARD"
+
+	// SUBSTRUCTURE represents substructure distance
+	SUBSTRUCTURE MetricType = "SUBSTRUCTURE"
+
+	// SUPERSTRUCTURE represents superstructure distance
+	SUPERSTRUCTURE MetricType = "SUPERSTRUCTURE"
 )

pkg/util/metric/similarity_corelation_test.go

@@ -43,6 +43,14 @@ func TestPositivelyRelated(t *testing.T) {
 			JACCARD,
 			false,
 		},
+		{
+			SUBSTRUCTURE,
+			false,
+		},
+		{
+			SUPERSTRUCTURE,
+			false,
+		},
 	}
 
 	for idx := range cases {

tests/python_client/common/common_type.py

@@ -239,7 +239,7 @@ delete_support = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]
 ivf = ["FLAT", "IVF_FLAT", "IVF_SQ8", "IVF_PQ"]
 skip_pq = ["IVF_PQ"]
 float_metrics = ["L2", "IP", "COSINE"]
-binary_metrics = ["JACCARD", "HAMMING"]
+binary_metrics = ["JACCARD", "HAMMING", "SUBSTRUCTURE", "SUPERSTRUCTURE"]
 structure_metrics = ["SUBSTRUCTURE", "SUPERSTRUCTURE"]
 all_scalar_data_types = ['int8', 'int16', 'int32', 'int64', 'float', 'double', 'bool', 'varchar']
 

tests/python_client/utils/util_pymilvus.py

@@ -75,7 +75,7 @@ def skip_pq():
 
 
 def binary_metrics():
-    return ["JACCARD", "HAMMING"]
+    return ["JACCARD", "HAMMING", "SUBSTRUCTURE", "SUPERSTRUCTURE"]
 
 
 def structure_metrics():