Fix self-loop detection to use the Definition 6 self-relation

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

src/split_miner/refined_dfg.py

@@ -73,11 +73,6 @@ class RefinedDirectlyFollowsGraph:
     ) -> None:
         """Build the refined DFG from lifecycle traces.
 
-        For each trace, scan for end events.  After each
-        end event, collect all start events that occur
-        before the next end event.  These form the
-        directly-follows pairs per Definition 6.
-
         :param traces: The event log.
         """
         for trace, count in traces.items():
@@ -98,17 +93,6 @@ class RefinedDirectlyFollowsGraph:
                 if lifecycle == "end":
                     self.__sinks.add(node)
                     break
-            # Detect self-loops: activity with multiple
-            # complete lifecycles in a trace.
-            end_counts: dict[Node, int] = {}
-            for node, lifecycle in trace:
-                if lifecycle == "end":
-                    end_counts[node] = (
-                        end_counts.get(node, 0) + 1
-                    )
-            for node, cnt in end_counts.items():
-                if cnt > 1:
-                    self.__self_loops.add(node)
             # Definition 6: ax ->r ay iff ay starts
             # after ax ends with no other end event
             # between.
@@ -121,12 +105,8 @@ class RefinedDirectlyFollowsGraph:
     ) -> None:
         """Scan a single trace for refined DF relations.
 
-        Walk through events.  When we see an end event
-        for activity ax, record ax as a "pending source".
-        When we see a start event for ay, create edges
-        from all pending sources to ay.  When we see
-        another end event, clear all pending sources
-        (since the new end event is "between").
+        A self-relation ax ->r ax is recorded as a
+        self-loop instead of an edge.
 
         :param trace: The lifecycle trace.
         :param count: The trace frequency.
@@ -144,14 +124,14 @@ class RefinedDirectlyFollowsGraph:
                 # All pending sources directly-follow
                 # to this activity.
                 for src in pending:
-                    if src != node:
+                    if src == node:
+                        self.__self_loops.add(node)
+                        continue
                     pair: tuple[Node, Node] = (
                         src, node
                     )
                     self.__df_freq[pair] = (
-                            self.__df_freq.get(
-                                pair, 0
-                            )
+                        self.__df_freq.get(pair, 0)
                         + count
                     )
 
@@ -190,8 +170,11 @@ class RefinedDirectlyFollowsGraph:
     def self_loops(self) -> set[Node]:
         """The set of self-loop nodes.
 
-        An activity is a self-loop if it completes
-        (has an end event) more than once in any trace.
+        An activity a is a self-loop if a ->r a holds per
+        Definition 6, that is, in some trace an end event
+        of a is followed by a start event of a with no
+        other end event in between.  Such self-relations
+        are excluded from :attr:`edges`.
 
         :return: The self-loop nodes.
         """

tests/test_refined_dfg.py

@@ -455,15 +455,16 @@ class TestForkAfterEnd(unittest.TestCase):
 class TestSelfLoop(unittest.TestCase):
     """Tests self-loop detection from lifecycle traces.
 
-    An activity that completes more than once in a trace
-    is a self-loop.
+    An activity a is a self-loop iff a ->r a holds per
+    Definition 6, that is, an end event of a is followed by
+    a start event of a with no other end event in between.
     """
 
     def setUp(self) -> None:
-        """Set up trace with self-loop: A repeats.
+        """Set up trace with self-loop: A repeats at once.
 
-        Trace: As Ae Bs Be As Ae Cs Ce
-        Activity A completes twice.
+        Trace: As Ae As Ae Bs Be Cs Ce
+        A ends and immediately starts again, so A ->r A.
 
         :return: None.
         """
@@ -477,8 +478,8 @@ class TestSelfLoop(unittest.TestCase):
             tuple[tuple[Task, str], ...], int
         ] = {
             ((a, S), (a, E),
-             (b, S), (b, E),
              (a, S), (a, E),
+             (b, S), (b, E),
              (c, S), (c, E)): 1,
         }
         self.__dfg: RefinedDirectlyFollowsGraph = (
@@ -505,6 +506,27 @@ class TestSelfLoop(unittest.TestCase):
             self.__dfg.self_loops,
         )
 
+    def test_no_self_edge(self) -> None:
+        """The self-loop edge A->A is not in the edges.
+
+        :return: None.
+        """
+        a: Task = self.__tasks["A"]
+        self.assertNotIn((a, a), self.__dfg.edges)
+
+    def test_other_edges(self) -> None:
+        """The other edges are A->B and B->C.
+
+        :return: None.
+        """
+        a: Task = self.__tasks["A"]
+        b: Task = self.__tasks["B"]
+        c: Task = self.__tasks["C"]
+        self.assertEqual(
+            self.__dfg.edges,
+            {(a, b), (b, c)},
+        )
+
     def test_no_self_loop_in_paper(self) -> None:
         """Paper example has no self-loops.
 
@@ -514,5 +536,100 @@ class TestSelfLoop(unittest.TestCase):
         self.assertEqual(dfg.self_loops, set())
 
 
+class TestSelfLoopWithoutRepeatedEnd(unittest.TestCase):
+    """Tests a self-loop with a single end event.
+
+    Definition 6 only requires an end event of a followed by
+    a start event of a, not a second end event of a.
+    """
+
+    def setUp(self) -> None:
+        """Set up trace: As Ae As Bs Be.
+
+        A ends, then A starts again with no end event in
+        between, so A ->r A.
+
+        :return: None.
+        """
+        self.__tasks: dict[str, Task] = (
+            _make_tasks("A", "B")
+        )
+        a: Task = self.__tasks["A"]
+        b: Task = self.__tasks["B"]
+        traces: dict[
+            tuple[tuple[Task, str], ...], int
+        ] = {
+            ((a, S), (a, E), (a, S),
+             (b, S), (b, E)): 1,
+        }
+        self.__dfg: RefinedDirectlyFollowsGraph = (
+            RefinedDirectlyFollowsGraph(traces)
+        )
+
+    def test_a_is_self_loop(self) -> None:
+        """A is a self-loop even though it ends once.
+
+        :return: None.
+        """
+        self.assertIn(
+            self.__tasks["A"],
+            self.__dfg.self_loops,
+        )
+
+
+class TestTwoLoopIsNotSelfLoop(unittest.TestCase):
+    """Tests that a 2-loop is not mistaken for a self-loop.
+
+    An activity repeating after another activity completes
+    forms a 2-loop, not a self-loop.
+    """
+
+    def setUp(self) -> None:
+        """Set up trace: As Ae Bs Be As Ae.
+
+        B ends between the end of A and the next start of
+        A, so A ->r A does not hold.
+
+        :return: None.
+        """
+        self.__tasks: dict[str, Task] = (
+            _make_tasks("A", "B")
+        )
+        a: Task = self.__tasks["A"]
+        b: Task = self.__tasks["B"]
+        traces: dict[
+            tuple[tuple[Task, str], ...], int
+        ] = {
+            ((a, S), (a, E),
+             (b, S), (b, E),
+             (a, S), (a, E)): 1,
+        }
+        self.__dfg: RefinedDirectlyFollowsGraph = (
+            RefinedDirectlyFollowsGraph(traces)
+        )
+
+    def test_a_not_self_loop(self) -> None:
+        """A is not a self-loop.
+
+        :return: None.
+        """
+        self.assertNotIn(
+            self.__tasks["A"],
+            self.__dfg.self_loops,
+        )
+
+    def test_two_loop_edges(self) -> None:
+        """The edges form the 2-loop A->B and B->A.
+
+        :return: None.
+        """
+        a: Task = self.__tasks["A"]
+        b: Task = self.__tasks["B"]
+        self.assertEqual(
+            self.__dfg.edges,
+            {(a, b), (b, a)},
+        )
+
+
 if __name__ == "__main__":
     unittest.main()