Merge pull request #13 from bio-ontology-research-group/owl2vec

🎨 ✨ ✅ Moved OWL2VecStar from develop di…

Author: ferzcam

docs/source/api/embedding/graph_based/index.rst

@@ -11,13 +11,21 @@ DL2Vec
    :undoc-members:
    :show-inheritance:
 
-
 Example of use
 ^^^^^^^^^^^^^^^
 An example of use can be found at :doc:`../../../tutorials/DL2Vec`
 
 
 
+OWL2Vec*
+--------
+
+.. automodule:: mowl.embeddings.graph_based.owl2vec_star.model
+   :members:
+   :undoc-members:
+   :show-inheritance:
+
+
 Translational Embeddings
 -------------------------
 

mowl/develop/owl2vec/model.py

@@ -29,12 +29,18 @@ class DL2Vec(Model):
     :type outfile: str
     :param bidirectional_taxonomy: If true, the ontology projection into a graph will add inverse edges per each subclass axiom
     :type bidirectional_taxonomy: bool
+    :param walking_method: Method for generating the walks. Choices are: deepwalk (default), node2vec, walkrdfowl.
+    :type walking_method: str
     :param walk_length: Length of the walk performed for each node
     :type walk_length: int
     :param num_walks: Number of walks performed per node
     :type num_walks: int
-    :param alpha: Probability of restart in the walking phase
+    :param alpha: Probability of restart in the walking phase. Applicable with DeepWalk
     :type alpha: float
+    :param p: Return hyperparameter. Default is 1. Applicable with Node2Vec
+    :type p: float
+    :param q: In-out hyperparameter. Default is 1. Applicable with Node2Vec.
+    :type q: float
     :param vector_size: Dimensionality of the word vectors. Same as :class:`gensim.models.word2vec.Word2Vec`
     :type vector_size: int
     :param wv_epochs: Number of epochs for the Word2Vec model

mowl/embeddings/graph_based/dl2vec/model.py

@@ -20,35 +20,39 @@
 
 logging.basicConfig(level=logging.INFO)
 
-class OWL2Vec(Model):
-
+class OWL2VecStar(Model):
     '''
     :param dataset: Dataset composed by training, validation and testing sets, each of which are in OWL format.
     :type dataset: :class:`mowl.datasets.base.Dataset`
     :param outfile: Path to save the final model
     :type outfile: str
     :param bidirectional_taxonomy: If true, the ontology projection into a graph will add inverse edges per each subclass axiom
     :type bidirectional_taxonomy: bool
-    :param include_literals: If true the graph will also include triples involving data property assertions and annotations. Default is False.
-    :type include_literals: bool
-    :param only_taxonomy: If true, the projection will only include subClass edges
+    :param only_taxonomy: If true, the ontology projection will consider only subclass axioms.
     :type only_taxonomy: bool
+    :param include_literals: If true, the graph will also include triples involving data property assertions and annotations.
+    :type include_literals: bool
+    :param walking_method: Method for generating the walks. Choices are: deepwalk (default), node2vec, walkrdfowl.
+    :type walking_method: str
     :param walk_length: Length of the walk performed for each node
     :type walk_length: int
     :param num_walks: Number of walks performed per node
     :type num_walks: int
-    :param alpha: Probability of restart in the walking phase
+    :param alpha: Probability of restart in the walking phase. Applicable with DeepWalk
     :type alpha: float
+    :param p: Return hyperparameter. Default is 1. Applicable with Node2Vec
+    :type p: float
+    :param q: In-out hyperparameter. Default is 1. Applicable with Node2Vec.
+    :type q: float
     :param vector_size: Dimensionality of the word vectors. Same as :class:`gensim.models.word2vec.Word2Vec`
     :type vector_size: int
     :param wv_epochs: Number of epochs for the Word2Vec model
     :type wv_epochs: int
     :param window: Maximum distance between the current and predicted word within a sentence. Same as :class:`gensim.models.word2vec.Word2Vec`
     :type window: int
-    :param num_procs: Number of threads to use for the random walks and the Word2Vec model.
-    :type num_procs: int
+    :param workers: Number of threads to use for the random walks and the Word2Vec model.
+    :type workers: int
     '''
-
 
     def __init__(self, dataset, outfile, bidirectional_taxonomy=False, include_literals = False, only_taxonomy = False, walking_method = "deepwalk", walk_length = 30, wv_epochs = 10, alpha = 0, num_walks = 100, vector_size = 100, window = 5, workers = 1, p = 1, q=1):
 

mowl/embeddings/graph_based/owl2vec/model.py

@@ -2,7 +2,7 @@
 from mowl.graph.taxonomyRels.model import TaxonomyWithRelsParser
 from mowl.graph.dl2vec.model import DL2VecParser
 
-from mowl.graph.owl2vec_star.model import OWL2VecParser
+from mowl.graph.owl2vec_star.model import OWL2VecStarParser
 
 def parser_factory(method_name, dataset, bidirectional_taxonomy, include_literals = False, only_taxonomy = False): 
     methods = [
@@ -19,6 +19,6 @@ def parser_factory(method_name, dataset, bidirectional_taxonomy, include_literal
     elif method_name == "dl2vec":
         return DL2VecParser(dataset, bidirectional_taxonomy= bidirectional_taxonomy)
     elif method_name == "owl2vec_star":
-        return OWL2VecParser(dataset, bidirectional_taxonomy=bidirectional_taxonomy, include_literals = include_literals, only_taxonomy = only_taxonomy)
+        return OWL2VecStarParser(dataset, bidirectional_taxonomy=bidirectional_taxonomy, include_literals = include_literals, only_taxonomy = only_taxonomy)
     else:
         raise Exception(f"Graph generation method unrecognized. Recognized methods are: {methods}")