'''
PM4Py – A Process Mining Library for Python
Copyright (C) 2024 Process Intelligence Solutions UG (haftungsbeschränkt)
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU Affero General Public License as
published by the Free Software Foundation, either version 3 of the
License, or any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License
along with this program. If not, see this software project's root or
visit <https://www.gnu.org/licenses/>.
Website: https://processintelligence.solutions
Contact: info@processintelligence.solutions
'''
from pm4py.objects.ocel.obj import OCEL
from pm4py.algo.discovery.ocel.ocdfg.variants import classic as ocdfg_discovery
from pm4py.algo.discovery.inductive import algorithm as inductive_miner
from enum import Enum
from pm4py.util import exec_utils
from pm4py.objects.ocel import constants as ocel_constants
from collections import Counter
from typing import Optional, Dict, Any
from pm4py.objects.dfg.obj import DFG
from pm4py.objects.conversion.process_tree import converter as tree_converter
from pm4py.algo.conformance.tokenreplay import algorithm as token_based_replay
from pm4py.objects.ocel.util import flattening
from copy import copy
[docs]
class Parameters(Enum):
EVENT_ACTIVITY = ocel_constants.PARAM_EVENT_ACTIVITY
OBJECT_TYPE = ocel_constants.PARAM_OBJECT_TYPE
INDUCTIVE_MINER_VARIANT = "inductive_miner_variant"
DOUBLE_ARC_THRESHOLD = "double_arc_threshold"
DIAGNOSTICS_WITH_TBR = "diagnostics_with_token_based_replay"
[docs]
def apply(
ocel: OCEL, parameters: Optional[Dict[Any, Any]] = None
) -> Dict[str, Any]:
"""
Discovers an object-centric Petri net (without annotation) from the given object-centric event log,
using the Inductive Miner as process discovery algorithm.
Reference paper: van der Aalst, Wil MP, and Alessandro Berti. "Discovering object-centric Petri nets." Fundamenta informaticae 175.1-4 (2020): 1-40.
Parameters
-----------------
ocel
Object-centric event log
parameters
Parameters of the algorithm, including:
- Parameters.EVENT_ACTIVITY => the activity attribute to be used
- Parameters.OBJECT_TYPE => the object type attribute to be used
- Parameters.DOUBLE_ARC_THRESHOLD => the threshold for the attribution of the "double arc", as
described in the paper.
- Parameters.DIAGNOSTICS_WITH_TBR => performs token-based replay and stores the result in the return dict
Returns
-----------------
ocpn
Object-centric Petri net model, presented as a dictionary of properties:
- activities: complete set of activities derived from the object-centric event log
- object_types: complete set of object types derived from the object-centric event log
- edges: dictionary connecting each object type to a set of directly-followed arcs between activities (expressed as tuples,
e.g., (act1, act2)). Every pair of activities is linked to some sets:
- event_pairs: the tuples of event identifiers where the directly-follows arc occurs
- total_objects: set of tuples containing two event and one object identifier, uniquely identifying an
occurrence of the arc.
- activities_indep: dictionary linking each activity, regardless of the object type, to some sets:
- events: the event identifiers where the activity occurs
- unique_objects: the object identifiers where the activity occurs
- total_objects: the tuples of event and object identifiers where the activity occurs.
- activities_ot: dictionary linking each object type to another dictionary, where the activities are linked to some sets:
- events: the event identifiers where the activity occurs (with at least one object of the given object type)
- unique_objects: the object identifiers of the given object type where the activity occurs
- total_objects: the tuples of event and object identifiers where the activity occurs.
- start_activities: dictionary linking each object type to another dictionary, where the start activities
of the given object type are linked to some sets:
- events: the event identifiers where the start activity occurs (with at least one object of the given object type)
- unique_objects: the object identifiers of the given object type where the start activity occurs
- total_objects: the tuples of event and object identifiers where the start activity occurs.
- end_activities: dictionary linking each object type to another dictionary, where the end activities
of the given object type are linked to some sets:
- events: the event identifiers where the end activity occurs (with at least one object of the given object type)
- unique_objects: the object identifiers of the given object type where the end activity occurs
- total_objects: the tuples of event and object identifiers where the end activity occurs.
- petri_nets: the accepted Petri nets (Petri net + initial marking + final marking) discovered by the process discovery algorithm
- double_arcs_on_activity: dictionary linking each object type to another dictionary, where each arc of the Petri net
is linked to a boolean (True if it is a double arc)
- tbr_results: the results of the token-based replay operation (if required)
"""
if parameters is None:
parameters = {}
double_arc_threshold = exec_utils.get_param_value(
Parameters.DOUBLE_ARC_THRESHOLD, parameters, 0.8
)
inductive_miner_variant = exec_utils.get_param_value(
Parameters.INDUCTIVE_MINER_VARIANT, parameters, "im"
)
diagnostics_with_tbr = exec_utils.get_param_value(
Parameters.DIAGNOSTICS_WITH_TBR, parameters, False
)
ocdfg_parameters = copy(parameters)
ocdfg_parameters["compute_edges_performance"] = False
ocpn = ocdfg_discovery.apply(ocel, parameters=ocdfg_parameters)
petri_nets = {}
double_arcs_on_activity = {}
tbr_results = {}
for ot in ocpn["object_types"]:
activities_eo = ocpn["activities_ot"]["total_objects"][ot]
start_activities = {
x: len(y)
for x, y in ocpn["start_activities"]["events"][ot].items()
}
end_activities = {
x: len(y) for x, y in ocpn["end_activities"]["events"][ot].items()
}
dfg = {}
if ot in ocpn["edges"]["event_couples"]:
dfg = {
x: len(y)
for x, y in ocpn["edges"]["event_couples"][ot].items()
}
is_activity_double = {}
for act in activities_eo:
ev_obj_count = Counter([x[0] for x in activities_eo[act]])
this_single_amount = 0
for y in ev_obj_count.values():
if y == 1:
this_single_amount += 1
this_single_amount = this_single_amount / len(ev_obj_count)
if this_single_amount <= double_arc_threshold:
is_activity_double[act] = True
else:
is_activity_double[act] = False
double_arcs_on_activity[ot] = is_activity_double
im_parameters = copy(parameters)
# disables the fallthroughs, as computing the model on a myriad of different object types
# could be really expensive
im_parameters["disable_fallthroughs"] = True
# for performance reasons, also disable the strict sequence cut (use
# the normal sequence cut)
im_parameters["disable_strict_sequence_cut"] = True
process_tree = None
flat_log = None
if inductive_miner_variant == "im" or diagnostics_with_tbr:
# do the flattening only if it is required
flat_log = flattening.flatten(ocel, ot, parameters=parameters)
if inductive_miner_variant == "imd":
obj = DFG()
obj._graph = Counter(dfg)
obj._start_activities = Counter(start_activities)
obj._end_activities = Counter(end_activities)
process_tree = inductive_miner.apply(
obj,
variant=inductive_miner.Variants.IMd,
parameters=im_parameters,
)
elif inductive_miner_variant == "im":
process_tree = inductive_miner.apply(
flat_log, parameters=im_parameters
)
petri_net = tree_converter.apply(process_tree, parameters=parameters)
if diagnostics_with_tbr:
tbr_parameters = copy(parameters)
tbr_parameters["enable_pltr_fitness"] = True
tbr_parameters["show_progress_bar"] = False
(
replayed_traces,
place_fitness_per_trace,
transition_fitness_per_trace,
notexisting_activities_in_model,
) = token_based_replay.apply(
flat_log,
petri_net[0],
petri_net[1],
petri_net[2],
parameters=tbr_parameters,
)
place_diagnostics = {
place: {"m": 0, "r": 0, "c": 0, "p": 0}
for place in place_fitness_per_trace
}
trans_count = {trans: 0 for trans in petri_net[0].transitions}
# computes the missing, remaining, consumed, and produced tokens
# per place.
for place, res in place_fitness_per_trace.items():
place_diagnostics[place]["m"] += res["m"]
place_diagnostics[place]["r"] += res["r"]
place_diagnostics[place]["c"] += res["c"]
place_diagnostics[place]["p"] += res["p"]
# counts the number of times a transition has been fired during the
# replay.
for trace in replayed_traces:
for trans in trace["activated_transitions"]:
trans_count[trans] += 1
tbr_results[ot] = (place_diagnostics, trans_count)
petri_nets[ot] = petri_net
ocpn["petri_nets"] = petri_nets
ocpn["double_arcs_on_activity"] = double_arcs_on_activity
ocpn["tbr_results"] = tbr_results
return ocpn
