Merge branch 'master' of ssh://the-cake-is-a-lie.net:20022/jonas/NLP-LAB

This commit is contained in:
Jonas Weinz 2018-07-16 11:55:23 +02:00
commit 1baeaa865a
5 changed files with 57 additions and 46 deletions

View File

@ -15,6 +15,7 @@ from nltk.corpus import wordnet
import math
import pprint
from gensim.models import Word2Vec, KeyedVectors
# # Naive Approach
table = pd.read_csv('../Tools/emoji_descriptions.csv')
@ -29,25 +30,25 @@ for index, row in table.iterrows():
# Helper functions
#######################
def stemming(messages):
stemmed_messages = []
def stemming(message):
ps = PorterStemmer()
for m in messages:
words = word_tokenize(m)
sm = []
for w in words:
sm.append(ps.stem(w))
m = (" ").join(sm)
stemmed_messages.append(m)
return stemmed_messages
words = word_tokenize(message)
sm = []
for w in words:
sm.append(ps.stem(w))
stemmed_message = (" ").join(sm)
return stemmed_message
# * compare words to emoji descriptions
def evaluate_sentence(sentence, description_key = 'description', lang = 'eng', emojis_to_consider="all"):
def evaluate_sentence(sentence, description_key = 'description', lang = 'eng', emojis_to_consider="all", stem=True):
# assumes there is a trained w2v model stored in the same directory!
wv = KeyedVectors.load("word2vec.model", mmap='r')
if (stem):
sentence = stemming(sentence)
tokenized_sentence = word_tokenize(sentence)
n = len(tokenized_sentence)
l = table.shape[0]
matrix_list = []
for index in tableDict.keys():
@ -57,20 +58,11 @@ def evaluate_sentence(sentence, description_key = 'description', lang = 'eng', e
mat = np.zeros(shape=(m,n))
for i in range(len(emoji_tokens)):
for j in range(len(tokenized_sentence)):
syn1 = wordnet.synsets(emoji_tokens[i],lang=lang)
if len(syn1) == 0:
continue
w1 = syn1[0]
#print(j, tokenized_sentence)
syn2 = wordnet.synsets(tokenized_sentence[j], lang=lang)
if len(syn2) == 0:
continue
w2 = syn2[0]
val = w1.wup_similarity(w2)
if val is None:
try:
val = wv.similarity(emoji_tokens[i], tokenized_sentence[j])
except KeyError:
continue
mat[i,j] = val
#print(row['character'], mat)
matrix_list.append(mat)
return matrix_list
@ -83,10 +75,13 @@ def evaluate_sentence(sentence, description_key = 'description', lang = 'eng', e
# load and preprocess data
# emojis_to_consider can be either a list or "all"
def prepareData(stemming=False):
if(stemming):
def prepareData(stem=True, lower=True):
if(stem):
for index in tableDict.keys():
tableDict[index][1] = stemming(tableDict[index][1])
if(lower):
for index in tableDict.keys():
tableDict[index][1] = tableDict[index][1].lower()
#collect the emojis
lookup = {}

Binary file not shown.

View File

@ -40,7 +40,8 @@
"import simple_twitter_learning as stl\n",
"import glob\n",
"import sys\n",
"from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer, HashingVectorizer"
"from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer, HashingVectorizer\n",
"import pickle"
]
},
{
@ -144,7 +145,7 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "d304cda50752491da1637b292a9367e8",
"model_id": "d00ff918ad4d473499b1e91d4dcb8702",
"version_major": 2,
"version_minor": 0
},
@ -173,7 +174,8 @@
" ],\n",
" [\n",
" (widgets.BoundedIntText(value=-1,disabled=True,min=-1, max=10), \"k_means_cluster\"),\n",
" (widgets.BoundedIntText(value=20,disabled=True,min=-1, max=100), \"n_top_emojis\")\n",
" (widgets.BoundedIntText(value=20,disabled=True,min=-1, max=100), \"n_top_emojis\"),\n",
" (widgets.Dropdown(options=[\"latest\", \"mean\"], value=\"latest\"), \"label_criteria\")\n",
" ],\n",
" [\n",
" (widgets.Button(disabled=True),\"load_data\")\n",
@ -205,6 +207,7 @@
" ],\n",
" [\n",
" (widgets.Checkbox(value=True),\"use_doc2vec\"),\n",
" (widgets.Checkbox(value=True),\"d2v_use_pretrained\"),\n",
" (widgets.IntText(value=100),\"d2v_size\"),\n",
" (widgets.IntText(value=8), \"d2v_window\"),\n",
" (widgets.IntSlider(value=5, min=0, max=32), \"d2v_min_count\")\n",
@ -444,13 +447,16 @@
" if lemm_and_stemm:\n",
" p_s = progress_indicator(\"stemming progress\")\n",
" \n",
" emoji_mean = shown_widgets[\"label_criteria\"].value == \"mean\"\n",
" \n",
" sdm = stl.sample_data_manager.generate_and_read(path=shown_widgets[\"root_path\"].value,\n",
" n_top_emojis=shown_widgets[\"n_top_emojis\"].value,\n",
" file_range=range(r[0], r[1]),\n",
" n_kmeans_cluster=shown_widgets[\"k_means_cluster\"].value,\n",
" read_progress_callback=p_r.update,\n",
" stem_progress_callback=p_s.update if lemm_and_stemm else None,\n",
" apply_stemming = lemm_and_stemm)\n",
" apply_stemming = lemm_and_stemm,\n",
" emoji_mean=emoji_mean)\n",
" shown_widgets[\"batch_size\"].max = len(sdm.labels)\n",
" \n",
" \n",
@ -558,9 +564,12 @@
" # creating the vectorizer\n",
" vectorizer = None\n",
" if shown_widgets[\"use_doc2vec\"].value:\n",
" vectorizer = stl.skd2v.Doc2VecTransformer(size=shown_widgets[\"d2v_size\"].value,\n",
" window=shown_widgets[\"d2v_window\"].value,\n",
" min_count=shown_widgets[\"d2v_min_count\"].value)\n",
" if shown_widgets[\"d2v_use_pretrained\"].value:\n",
" vectorizer = pickle.load( open( \"doc2VecModel.p\", \"rb\" ) )\n",
" else:\n",
" vectorizer = stl.skd2v.Doc2VecTransformer(size=shown_widgets[\"d2v_size\"].value,\n",
" window=shown_widgets[\"d2v_window\"].value,\n",
" min_count=shown_widgets[\"d2v_min_count\"].value)\n",
" else:\n",
" vectorizer=TfidfVectorizer(stop_words='english')\n",
" \n",

Binary file not shown.

View File

@ -28,6 +28,8 @@ nltk.download('punkt')
nltk.download('averaged_perceptron_tagger')
nltk.download('wordnet')
from keras import losses
# check whether the display function exists:
try:
display
@ -52,7 +54,6 @@ def sent2emoji(sent_arr, custom_target_emojis=None, only_emoticons=True):
# In[3]:
SINGLE_LABEL = True
@ -161,7 +162,7 @@ def batch_lemm(sentences):
class sample_data_manager(object):
@staticmethod
def generate_and_read(path:str, only_emoticons=True, apply_stemming=True, n_top_emojis=-1, file_range=None, n_kmeans_cluster=-1, read_progress_callback=None, stem_progress_callback=None):
def generate_and_read(path:str, only_emoticons=True, apply_stemming=True, n_top_emojis=-1, file_range=None, n_kmeans_cluster=-1, read_progress_callback=None, stem_progress_callback=None, emoji_mean=False):
"""
generate, read and process train data in one step.
@ -175,7 +176,7 @@ class sample_data_manager(object):
@return: sample_data_manager object
"""
sdm = sample_data_manager(path)
sdm.read_files(file_index_range=range(sdm.n_files) if file_range is None else file_range, only_emoticons=only_emoticons, progress_callback=read_progress_callback)
sdm.read_files(file_index_range=range(sdm.n_files) if file_range is None else file_range, only_emoticons=only_emoticons, progress_callback=read_progress_callback, emoji_mean=emoji_mean)
if apply_stemming:
sdm.apply_stemming_and_lemmatization(progress_callback=stem_progress_callback)
@ -239,7 +240,7 @@ class sample_data_manager(object):
# so far filtering for the latest emoji. TODO: maybe there are also better approaches
labels_i = emoji2sent([latest(e) for e in emojis_i], only_emoticons=only_emoticons )
else:
labels_i = np.array([np.mean(emoji2sent(e, only_emoticons=only_emoticons), axis=0) for e in emojis_i])
labels_i = np.array([np.mean(emoji2sent(e, only_emoticons=only_emoticons), axis=0).tolist() for e in emojis_i])
# and filter out all samples we have no label for:
wrong_labels = np.isnan(np.linalg.norm(labels_i, axis=1))
@ -431,7 +432,7 @@ class pipeline_manager(object):
return pm
@staticmethod
def create_keras_pipeline_with_vectorizer(vectorizer, layers, sdm:sample_data_manager, loss=None, optimizer=None):
def create_keras_pipeline_with_vectorizer(vectorizer, layers, sdm:sample_data_manager, loss=None, optimizer=None, fit_vectorizer=True):
'''
creates pipeline with vectorizer and keras classifier
@ -447,11 +448,12 @@ class pipeline_manager(object):
from keras.models import Sequential
from keras.layers import Dense
if sdm.X is None:
sdm.create_train_test_split()
vec_train = vectorizer.fit_transform(sdm.X)
vec_test = vectorizer.transform(sdm.Xt)
if fit_vectorizer:
if sdm.X is None:
sdm.create_train_test_split()
vec_train = vectorizer.fit_transform(sdm.X)
vec_test = vectorizer.transform(sdm.Xt)
# creating keras model:
model=Sequential()
@ -578,7 +580,7 @@ class pipeline_manager(object):
"""fitting the pipeline"""
self.pipeline.fit(X,y)
def predict(self,X, use_stemming=True, use_lemmatization=True):
def predict(self,X, use_stemming=False, use_lemmatization=False):
"""predict"""
if use_stemming:
X = np.array(batch_stem(X))
@ -608,7 +610,7 @@ class trainer(object):
self.sdm = sdm
self.pm = pm
def fit(self, max_size=10000, disabled_fit_steps=['vectorizer'], keras_batch_fitting_layer=['keras_model'], batch_size=None, n_epochs=1, progress_callback=None):
def fit(self, max_size=1000000, disabled_fit_steps=['vectorizer'], keras_batch_fitting_layer=['keras_model'], batch_size=None, n_epochs=1, progress_callback=None):
"""
fitting data in the pipeline. Because we don't want to refit the vectorizer, the pipeline models containing the vectorizer have to be named explicitly
@ -641,7 +643,12 @@ class trainer(object):
named_steps[k].fit = lambda X, y: named_steps[k].train_on_batch(to_dense_if_sparse(X), y) # ← why has keras no sparse support on batch progressing!?!?!
if batch_size is None:
self.pm.fit(X = self.sdm.X[:max_size], y = self.sdm.y[:max_size])
for e in range(n_epochs):
print("epoch", e)
self.pm.fit(X = self.sdm.X[:max_size], y = self.sdm.y[:max_size])
pred, yt = self.test()
mean_squared_error = ((pred - yt)**2).mean(axis=0)
print("#" + str(e) + ": validation loss: ", mean_squared_error, "scalar: ", np.mean(mean_squared_error))
else:
n = len(self.sdm.X) // batch_size
for i in range(n_epochs):