def store_to_file (self, path) :
transcripts.printf_information ('Storing data to file [%s]...', path)
if os.path.exists (path) :
transcripts.printf_warning ('File does exist; overwriting.')
stream = file (path, 'wb', 512 * 1024)
self.store_to_stream (stream)
stream.close ()
def execute_training_1 (self, pattern) :
best_matched_prototype = None
best_matched_activation = None
matched_prototypes = list ()
unmatched_prototypes = list ()
for prototype in self.prototypes :
if self.match (prototype, pattern) > self.vigilance:
activation = self.activation (prototype, pattern)
if best_matched_activation is None :
best_matched_prototype = prototype
best_matched_activation = activation
elif best_matched_activation < activation :
matched_prototypes.append (best_matched_prototype)
best_matched_prototype = prototype
best_matched_activation = activation
else :
matched_prototypes.append (prototype)
else :
unmatched_prototypes.append (prototype)
if best_matched_prototype is None :
best_matched_prototype = pattern
adjusted_prototypes = list ()
adjusted_prototypes.append (self.adjust (best_matched_prototype, pattern, self.learning))
for prototype in matched_prototypes :
adjusted_prototypes.append (prototype)
for prototype in unmatched_prototypes :
adjusted_prototypes.append (prototype)
new_prototypes = list ()
for prototype in adjusted_prototypes :
if vectors.magnitude (prototype) <= 0.000000001 :
transcripts.printf_warning ('Prototype vector is almost 0; pruning.')
else :
new_prototypes.append (prototype)
self.prototypes = new_prototypes
self.prototype_stepper.update (len (self.prototypes))
def load_from_file (self, path) :
transcripts.printf_information ('Loading data from file [%s]...', path)
if not os.path.exists (path) :
transcripts.printf_warning ('File does not exist; aborting.')
return
stream = file (path, 'rb', 512 * 1024)
self.load_from_stream (stream)
stream.close ()
def store (self) :
transcripts.printf_information ('Storing data to stream [%s]...', self.data_path)
stream = streams.open (self.data_path, 'o')
if stream is None :
transcripts.printf_warning ('Stream does not exist; aborting.')
pickle.dump (self.document_identifiers, stream)
pickle.dump (self.document_titles, stream)
pickle.dump (self.document_contents, stream)
pickle.dump (self.document_vectors, stream)
stream.close ()
def load (self) :
transcripts.printf_information ('Loading data from stream [%s]...', self.data_path)
stream = streams.open (self.data_path, 'i')
if stream is None :
transcripts.printf_warning ('Stream does not exist; aborting.')
return
self.document_identifiers = pickle.load (stream)
self.document_titles = pickle.load (stream)
self.document_contents = pickle.load (stream)
self.document_vectors = pickle.load (stream)
stream.close ()
def execute_training (self, pattern) :
vector_length = len (pattern)
if self.vector_length is None :
self.vector_length = vector_length
else :
if self.vector_length != vector_length :
raise Exception ()
if vectors.magnitude (pattern) <= 0.000000001 :
transcripts.printf_warning ('Pattern vector is almost 0; ignoring.')
return
self.training_stepper.increment ()
self.execute_training_1 (pattern)
def dump(map, dumper):
id = map["id"][0]
if "topic" in map:
topic = map["topic"][0]
else:
transcripts.printf_warning("Missing topic(s) from article with id [%s]; assuming unknown.", id)
topic = "{unknown}"
if "title" in map:
title = map["title"][0]
else:
transcripts.printf_error("Missing title from article with id [%s]; assuming id.", id)
title = "(%s)" % (id)
if "content" in map:
content = "\n".join(map["content"])
else:
transcripts.printf_error("Missing content from article with id [%s]; ignoring.", id)
return
title = "[%s] %s" % (topic, title)
dumper.dump(title, content)