def _read_mapping(self, filename):
pairs = []
with open(filename) as f:
for line in f.readlines():
acm_id, wiki_title = line.strip().split('\t')
pairs.append((acm_id, to_category_uri(wiki_title)))
return zip(*pairs)
def _compute_parents(self, node):
"""Returns and caches the parents (super-categories) for the given node.
The parents are fetched from Wikipedia or from the local cache.
Unlike self.get_parents, returns also the parents that are not in the
'subtree' of self._root.
"""
parents = self._parents.get(node)
if not parents:
# parents = self._wiki_graph.get_supercats(node)
self._parents[node] = set(
to_category_uri(parent)
for parent in self._relations.parents(node))
def _compute_children(self, node):
"""Returns and caches the children (subcategories) for the given node.
The children are fetched from Wikipedia or from the local cache.
"""
children = self._children.get(node)
if children is None:
# children = set(self._wiki_graph.get_subcats(node) or [])
children = set(
to_category_uri(child)
for child in self._relations.children(node))
self._children[node] = children
return children
def is_relevant(self, node):
return to_category_uri(node) in self._relevant
def _compute_children(self, node):
subcat_uris = self._wiki_graph.\
get_subcats(dbpedia.to_category_uri(node))
return sorted([dbpedia.to_title(uri) for uri in subcat_uris])
def is_parent(self, parent, child):
return to_category_uri(child) in self.children(parent)
def _to_acm(self, node):
return self._wiki2acm[to_category_uri(node)]
def contains(self, node):
if is_acm_id(node):
return node in self._acm2wiki
else:
return to_category_uri(node) in self._wiki2acm
if self.contains(child)
]
else:
return set()
def is_parent(self, parent, child):
return to_category_uri(child) in self.children(parent)
def is_ancestor(self, parent, child):
if self.is_parent(parent, child):
return True
return any([
self.is_ancestor(parents_child, child)
for parents_child in self.children(parent)
])
acm = AcmWrapper()
assert acm._acm2wiki['10002950'] == to_category_uri('Mathematics of computing')
assert acm.is_parent('World Wide Web', 'Web applications')
assert not acm.is_parent('Web applications', 'Social networks')
assert not acm.is_ancestor('World Wide Web', 'Social networks')
# import pprint
# pprint.pprint(
# [(to_title(acm._acm2wiki[parent]), to_title(acm._acm2wiki[child]))
# for parent in acm._children.keys()
# for child in acm._children[parent]
# if parent in acm._acm2wiki and child in acm._acm2wiki])
def is_class(self, node):
node = to_category_uri(node)
if self._remember_gt and node in self._ground_truth:
return self._ground_truth[node]
else:
return self._predict(to_features(self._features, [node]))[0]
