zer-schema 1.0.2

Schema inference and Fellegi-Sunter model registry for the zer entity-resolution library
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56

# Value-pattern sampling heuristics for SchemaInferrer.
#
# When the column name gives no signal, SchemaInferrer samples up to 50 non-null
# text values and evaluates each pattern below in order. The first pattern whose
# ALL conditions are satisfied is returned.
#
# Fields per pattern:
#   kind           , FieldKind to assign (same names as the Rust enum variants)
#   regex          , regex applied to each sample; set to "" to skip regex matching
#   threshold      , minimum fraction of samples that must match the regex (0.0–1.0)
#   unique_rate_min, optional lower bound on (unique values / total samples)
#   unique_rate_max, optional upper bound on (unique values / total samples)
#   avg_len_min    , optional lower bound on mean string length
#   avg_len_max    , optional upper bound on mean string length
#
# A pattern with regex = "" and threshold = 0.0 matches purely on statistical
# conditions (unique_rate_*, avg_len_*).
#
# The [fallback] section sets the kind returned when no pattern matches.

[[patterns]]
kind      = "Date"
regex     = '^\d{4}[-/]\d{2}[-/]\d{2}$'
threshold = 0.8

[[patterns]]
kind      = "Phone"
regex     = '^[\d\s\-\+\(\)]{7,15}$'
threshold = 0.8

[[patterns]]
kind      = "Numeric"
regex     = '^\d+(\.\d+)?$'
threshold = 0.9

[[patterns]]
kind            = "Id"
regex           = '^[A-Z0-9\-]{4,20}$'
threshold       = 0.8
unique_rate_min = 0.9

[[patterns]]
kind            = "Name"
regex           = ""
threshold       = 0.0
unique_rate_min = 0.7
avg_len_max     = 30.0

[[patterns]]
kind            = "Categorical"
regex           = ""
threshold       = 0.0
unique_rate_max = 0.2

[fallback]
default_kind = "FreeText"