def train_svm(documents, ntesting=500):
"""
:param documents- politeness-annotated training data
:type documents- list of dicts
each document must be preprocessed and
'sentences' and 'parses' and 'score' fields.
:param ntesting- number of docs to reserve for testing
:type ntesting- int
returns fitted SVC, which can be serialized using cPickle
"""
# Generate and persist list of unigrams, bigrams
PolitenessFeatureVectorizer.generate_bow_features(documents)
# For good luck
random.shuffle(documents)
testing = documents[-ntesting:]
documents = documents[:-ntesting]
# SAVE FOR NOW
cPickle.dump(testing, open("testing-data.p", 'w'))
X, y = documents2feature_vectors(documents)
Xtest, ytest = documents2feature_vectors(testing)
print "Fitting"
clf = svm.SVC(C=0.02, kernel='linear', probability=True)
clf.fit(X, y)
# Test
y_pred = clf.predict(Xtest)
print(classification_report(ytest, y_pred))
return clf
def train_svm(documents, ntesting=500):
"""
:param documents- politeness-annotated training data
:type documents- list of dicts
each document must be preprocessed and
'sentences' and 'parses' and 'score' fields.
:param ntesting- number of docs to reserve for testing
:type ntesting- int
returns fitted SVC, which can be serialized using cPickle
"""
# Generate and persist list of unigrams, bigrams
PolitenessFeatureVectorizer.generate_bow_features(documents)
# For good luck
random.shuffle(documents)
testing = documents[-ntesting:]
documents = documents[:-ntesting]
# SAVE FOR NOW
pickle.dump(testing, open("model_svm.pkl", 'wb'))
X, y = documents2feature_vectors(documents)
Xtest, ytest = documents2feature_vectors(testing)
print("Fitting")
clf = svm.SVC(C=0.02, kernel='linear', probability=True)
clf.fit(X, y)
# Test
y_pred = clf.predict(Xtest)
print(classification_report(ytest, y_pred))
return clf
def documents2feature_vectors(documents):
vectorizer = PolitenessFeatureVectorizer()
fks = False
X, y = [], []
for d in documents:
fs = vectorizer.features(d)
if not fks:
fks = sorted(fs.keys())
fv = [fs[f] for f in fks]
# If politeness score > 0.0,
# the doc is polite, class=1
l = 1 if d['score'] > 0.0 else 0
X.append(fv)
y.append(l)
X = csr_matrix(np.asarray(X))
y = np.asarray(y)
return X, y
def documents2feature_vectors(documents):
vectorizer = PolitenessFeatureVectorizer()
fks = False
X, y = [], []
for d in documents:
fs = vectorizer.features(d)
if not fks:
fks = sorted(fs.keys())
fv = [fs[f] for f in fks]
# If politeness score > 0.0,
# the doc is polite, class=1
l = 1 if d['score'] > 0.0 else 0
X.append(fv)
y.append(l)
X = csr_matrix(np.asarray(X))
y = np.asarray(y)
return X, y
X, y = [], []
for d in documents:
print(d)
fs = vectorizer.features(d)
if not fks:
fks = sorted(fs.keys())
fv = [fs[f] for f in fks]
# If politeness score > 0.0,
# the doc is polite, class=1
l = 1 if d['score'] > 0.0 else 0
X.append(fv)
y.append(l)
X = csr_matrix(np.asarray(X))
y = np.asarray(y)
return X, y
if __name__ == "__main__":
"""
Train a dummy model off our 4 sample request docs
"""
#from politeness.test_documents import TEST_DOCUMENTS
codings = pd.read_csv('Pol600withLabel.csv', index_col=0, parse_dates=True)
docs = codings['Request'].tolist()
documents = PolitenessFeatureVectorizer.preprocess(docs)
train_svm(documents, ntesting=1)
from politeness.features.vectorizer import PolitenessFeatureVectorizer
ret = PolitenessFeatureVectorizer.preprocess([
"what do you think was the purpose of this Fesselballon? if it wasn't military, what was the point in shooting it down?"
])
print(ret)
def index():
global login
if not login:
return redirect(url_for('login'))
cur = mysql.connection.cursor()
# cur.execute("""DROP TABLE IF EXISTS Feedback_Doc;""")
# cur.execute("""DROP TABLE IF EXISTS Feedback_Sentence;""")
# cur.execute("""DROP TABLE IF EXISTS Input;""")
cur.execute("""CREATE TABLE IF NOT EXISTS Input (
input_id INTEGER PRIMARY KEY AUTO_INCREMENT,
user_id TEXT,
message TEXT,
time_stamp TEXT
)""")
cur.execute("""CREATE TABLE IF NOT EXISTS Feedback_Doc (
input_id INTEGER PRIMARY KEY,
word_count INTEGER,
label TEXT,
impoliteness_score REAL,
politeness_score REAL,
FOREIGN KEY (input_id) REFERENCES Input(input_id)
)""")
cur.execute("""CREATE TABLE IF NOT EXISTS Feedback_Sentence (
id INTEGER PRIMARY KEY AUTO_INCREMENT,
input_id INTEGER,
sentence_content TEXT,
label TEXT,
impoliteness_score REAL,
politeness_score REAL,
strategy_count INTEGER,
strategies VARCHAR(255),
indices VARCHAR(255),
FOREIGN KEY (input_id) REFERENCES Input(input_id)
)""")
label_string = ""
input_text = ""
title = ""
strategies_set = set()
highlight_index_set = set()
strategies = []
strategies_all = []
if request.method == 'POST':
title = request.form['theme']
input_text = request.form['sentence']
# check for grammatical mistakes
grammar_check = api.check(input_text, api_url='https://languagetool.org/api/v2/', lang='en-US')
grammar_messages = grammar_check['matches']
grammar_corrections, split_input, wrong_words, impolite_words, replacements = [], [], [], [], {}
if len(grammar_messages) != 0:
for i in range(len(grammar_messages)):
# og_msg = grammar_messages[i]['context']['text']
og_msg = input_text
offset = grammar_messages[i]['offset']
grammar_corrections.append(grammar_messages[i]['message'])
wrong_words.append(og_msg[offset:offset+grammar_messages[i]['length']])
for repl in grammar_messages[i]['replacements']:
if i not in replacements:
replacements[i] = [repl['value']]
else:
replacements[i].append(repl['value'])
split_input = input_text.split()
### NEEDS TO BE CHANGED LATER...
num_corrections = str(len(replacements))
print(wrong_words)
print(replacements)
# Get politeness score for overall document
doc_res = score_text(input_text)
print("DOCUMENT POLITENESS:\n", doc_res)
label_string = doc_res[0]
# Get politeness score for each sentence in document
sentence_list = nltk.sent_tokenize(input_text)
sent_politeness_res = list()
impolite_sentence_indices = dict()
for i, sentence in enumerate(sentence_list):
## politeness score
res = score_text(sentence)
label, impolite_score, polite_score = res[0], res[1], res[2]
## strategies feedback
doc = PolitenessFeatureVectorizer.preprocess([sentence])[0]
strategies = get_feedback(doc)
for strat in strategies:
strategies_set.add(strat[0])
highlight_index_set.add(strat[1][0])
sent_politeness_res.append( (sentence, label, impolite_score, polite_score, strategies) )
print("PER SENTENCE POLITENESS\n", sent_politeness_res)
# print(sent_politeness_res[0][4])
# print(len(sent_politeness_res[0][4]))
strategies_all = sent_politeness_res[0][4]
now = datetime.datetime.now().strftime("%b %d %Y %H:%M:%S")
cur.execute("INSERT INTO Input (user_id, message, time_stamp) VALUES (%s, %s, %s)", (g.user, input_text, now))
cur.execute("SELECT input_id FROM Input WHERE time_stamp = %s", (now,))
input_id = cur.fetchone()[0]
cur.execute("INSERT INTO Feedback_Doc (input_id, word_count, label, impoliteness_score, politeness_score) VALUES (%s, %s, %s, %s, %s)",
(input_id, len(input_text.split()), doc_res[0], float(doc_res[1]), float(doc_res[2])))
for m in sent_politeness_res:
strategies = [i[0] for i in m[4]]
strategies_idx = [i[1] for i in m[4]]
cur.execute(
"INSERT INTO Feedback_Sentence (input_id, sentence_content, label, impoliteness_score, politeness_score, strategy_count, strategies, indices) VALUES (%s, %s, %s, %s, %s, %s, %s, %s)",
(input_id, m[0], m[1], float(m[2]), float(m[3]), len(m[4]), str(strategies), str(strategies_idx)))
mysql.connection.commit()
print(strategies)
# cur.execute("""SELECT * FROM Input""")
# print(cur.fetchall(), '\n')
# cur.execute("""SELECT * FROM Feedback_Doc""")
# print(cur.fetchall(), '\n')
# cur.execute("""SELECT * FROM Feedback_Sentence""")
# print(cur.fetchall(), '\n')
cur.close()
original_text = input_text
# return render_template('feedback.html', user_input=input_text, label_string=label_string, impoliteness_score=impoliteness_score, politeness_score=politeness_score, strategies=strategies, grammar_msg=grammar_corrections, repl=replacements, split_inputs=split_input, num_errors=num_corrections, mistakes=wrong_words, impolite_ind=impolite_indices, impolite_words=impolite_words)
return render_template('new_feedback.html',label_string = label_string, user_input = input_text, title = title,strategies_list = strategies_set, strategies = strategies_all, highlight_index = highlight_index_set)
