"""Learning problem in Ontolearn."""
import logging
from typing import Set, Optional, TYPE_CHECKING
if TYPE_CHECKING:
from ontolearn.knowledge_base import KnowledgeBase
from ontolearn.abstracts import AbstractLearningProblem, EncodedLearningProblem, EncodedPosNegLPStandardKind
from owlapy.owl_individual import OWLNamedIndividual
logger = logging.getLogger(__name__)
[docs]
class EncodedPosNegLPStandard(EncodedPosNegLPStandardKind):
"""Encoded learning problem standard.
Attributes:
kb_pos (set): Positive examples.
kb_neg (set): Negative examples.
kb_diff (set): kb_all - (kb_pos + kb_neg).
kb_all (set): All examples/ all individuals set.
"""
__slots__ = 'kb_pos', 'kb_neg', 'kb_diff', 'kb_all'
kb_pos: set
kb_neg: set
kb_diff: set
kb_all: set
def __init__(self, kb_pos, kb_neg, kb_diff, kb_all):
"""Create a new instance of EncodedPosNegLPStandard.
Args:
kb_pos (set): Positive examples.
kb_neg (set): Negative examples.
kb_diff (set): kb_all - (kb_pos + kb_neg).
kb_all (set): All examples/ all individuals set.
"""
self.kb_pos = kb_pos
self.kb_neg = kb_neg
self.kb_diff = kb_diff
self.kb_all = kb_all
[docs]
class PosNegLPStandard(AbstractLearningProblem):
"""Positive-Negative learning problem standard.
Attributes:
pos: Positive examples.
neg: Negative examples.
all: All examples.
"""
__slots__ = 'pos', 'neg', 'all'
def __init__(self,
pos: Set[OWLNamedIndividual],
neg: Set[OWLNamedIndividual],
all_instances: Optional[Set[OWLNamedIndividual]] = None):
"""
Determine the learning problem and initialize the search.
1) Convert the string representation of an individuals into the owlready2 representation.
2) Sample negative examples if necessary.
3) Initialize the root and search tree.
Args:
pos: Positive examples.
neg: Negative examples.
all_instances: All examples.
"""
assert isinstance(pos, set) and isinstance(neg, set)
self.pos = frozenset(pos)
self.neg = frozenset(neg)
if all_instances is None:
self.all = None
else:
self.all = frozenset(all_instances)
[docs]
def encode_kb(self, knowledge_base: 'KnowledgeBase') -> EncodedPosNegLPStandard:
return knowledge_base.encode_learning_problem(self)
[docs]
class EncodedPosNegUndLP(EncodedLearningProblem):
"""To be implemented."""
...
# XXX: TODO
[docs]
class PosNegUndLP(AbstractLearningProblem):
"""To be implemented."""
...
# XXX: TODO