Pfdtane nongeneralized

cli/cli.py

@@ -21,6 +21,7 @@ class Task(StrEnum):
 class Algorithm(StrEnum):
     pyro = auto()
     tane = auto()
+    pfdtane = auto()
     hyfd = auto()
     fd_mine = auto()
     dfd = auto()
@@ -131,15 +132,15 @@ class Algorithm(StrEnum):
 “Functional dependency discovery: an experimental evaluation of seven
 algorithms” paper by T. Papenbrock et al.
 
-Algorithms: PYRO, TANE, HYFD, FD_MINE, DFD, DEP_MINER, FDEP, FUN, FASTFDS, AID
+Algorithms: PYRO, TANE, PFDTANE, HYFD, FD_MINE, DFD, DEP_MINER, FDEP, FUN, FASTFDS, AID
 Default: HYFD
 '''
 AFD_HELP = '''Discover minimal non-trivial approximate functional
 dependencies. Approximate functional dependencies are defined in the
 “Efficient Discovery of Approximate Dependencies” paper by S. Kruse and
 F. Naumann.
 
-Algorithms: PYRO, TANE
+Algorithms: PYRO, TANE, PFDTANE
 Default: PYRO
 '''
 FD_VERIFICATION_HELP = '''Verify whether a given exact functional dependency
@@ -176,6 +177,11 @@ class Algorithm(StrEnum):
 Algorithm for Discovering Functional and Approximate Dependencies” by
 Y. Huntala et al.
 '''
+PFDTANE_HELP = '''A TANE-based algorithm for discovery of probabilistic
+functional dependencies. For more information, refer to “Functional Dependency
+Generation and Applications in pay-as-you-go data integration systems” by
+Daisy Zhe Wang et al.
+'''
 HYFD_HELP = '''A modern algorithm for discovery of exact functional
 dependencies. One of the most high-performance algorithms for this task. For
 more information, refer to “A Hybrid Approach to Functional Dependency
@@ -252,6 +258,7 @@ class Algorithm(StrEnum):
 ALGO_HELP_PAGES = {
     Algorithm.pyro: PYRO_HELP,
     Algorithm.tane: TANE_HELP,
+    Algorithm.pfdtane: PFDTANE_HELP,
     Algorithm.hyfd: HYFD_HELP,
     Algorithm.fd_mine: FD_MINE_HELP,
     Algorithm.dfd: DFD_HELP,
@@ -268,12 +275,12 @@ class Algorithm(StrEnum):
 TaskInfo = namedtuple('TaskInfo', ['algos', 'default'])
 
 TASK_INFO = {
-    Task.fd: TaskInfo([Algorithm.pyro, Algorithm.tane, Algorithm.hyfd,
+    Task.fd: TaskInfo([Algorithm.pyro, Algorithm.tane, Algorithm.pfdtane, Algorithm.hyfd,
                        Algorithm.fd_mine, Algorithm.dfd, Algorithm.dep_miner,
                        Algorithm.fdep, Algorithm.fun, Algorithm.fastfds,
                        Algorithm.aid],
                       Algorithm.hyfd),
-    Task.afd: TaskInfo([Algorithm.pyro, Algorithm.tane],
+    Task.afd: TaskInfo([Algorithm.pyro, Algorithm.tane, Algorithm.pfdtane],
                        Algorithm.pyro),
     Task.fd_verification: TaskInfo([Algorithm.naive_fd_verifier],
                                    Algorithm.naive_fd_verifier),
@@ -286,6 +293,7 @@ class Algorithm(StrEnum):
 ALGOS = {
     Algorithm.pyro: desbordante.Pyro,
     Algorithm.tane: desbordante.Tane,
+    Algorithm.pfdtane: desbordante.PFDTane,
     Algorithm.hyfd: desbordante.HyFD,
     Algorithm.fd_mine: desbordante.FdMine,
     Algorithm.dfd: desbordante.DFD,

src/core/algorithms/algorithm_types.h

@@ -9,7 +9,7 @@ namespace algos {
 using AlgorithmTypes =
         std::tuple<Depminer, DFD, FastFDs, FDep, Fd_mine, Pyro, Tane, FUN, hyfd::HyFD, Aid, Apriori,
                    metric::MetricVerifier, DataStats, fd_verifier::FDVerifier, HyUCC, PyroUCC,
-                   cfd::FDFirstAlgorithm, ACAlgorithm, UCCVerifier, Faida>;
+                   cfd::FDFirstAlgorithm, ACAlgorithm, UCCVerifier, Faida, PFDTane>;
 
 // clang-format off
 /* Enumeration of all supported non-pipeline algorithms. If you implement a new
@@ -27,6 +27,7 @@ BETTER_ENUM(AlgorithmType, char,
     fdmine,
     pyro,
     tane,
+    pfdtane,
     fun,
     hyfd,
     aidfd,

src/core/algorithms/algorithms.h

@@ -9,6 +9,7 @@
 #include "algorithms/fd/fdep/fdep.h"
 #include "algorithms/fd/fun/fun.h"
 #include "algorithms/fd/hyfd/hyfd.h"
+#include "algorithms/fd/pfdtane/pfdtane.h"
 #include "algorithms/fd/pyro/pyro.h"
 #include "algorithms/fd/tane/tane.h"
 #include "algorithms/statistics/data_stats.h"

src/core/algorithms/fd/pfdtane/enums.h

@@ -0,0 +1,9 @@
+#pragma once
+
+#include <enum.h>
+
+namespace algos {
+
+BETTER_ENUM(ErrorMeasure, char, per_tuple = 0, per_value)
+
+}