alabrashJr

#fname: the file name of binary file <google_w2v.bin>
#vocab: vocabulary dictionary

function load_bin_vec(fname, vocab)
pc(s)=return convert(Char,s[1])
word_vecs = Dict()
    open(fname, "r") do f
                @show header = readline(f)
                vocab_size, layer1_size = map(pf, split(header))
                @show binary_len = sizeof(Float32) * layer1_size

alabrashJr

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "import requests \n",
    "from bs4 import BeautifulSoup\n",

alabrashJr

#this script was written by Cigil and edit by aalabrash18@ku.edu.tr
import pandas as pd
from lxml import etree
import re
import os, sys, io, traceback, codecs
from bs4 import BeautifulSoup, Comment
UNESCAPE = True
import html as h
import glob

alabrashJr

Çoklu Etiket Metin Sınıflandırma

Sınıflandırma Yöntemleri

köpek midir?	hangi etkit ?	hangi etiketler?
Evet/Hayırikili	köpek/doğa/çimençok sınıflı	köpek,doğa,çimen****çoklu etiket

alabrashJr

def predict(text):
    print("text -> ",text)
    text_s=[stemmer.stem(w) for w in text.split()]
    print("text stemmed-> ",text_s)
    X=vectorizer.transform([text_s])
    X_pred=lp_classifier.predict(X).toarray()
    return [data.columns[i] for i,x in enumerate(X_pred[0]) if x==1]

text="araçım servise son getirmediğimde düzgün tamir edilmedi memnun değil"

alabrashJr

!pip install scikit-multilearn
!pip install scikit-learn
from sklearn.linear_model import LogisticRegression
from skmultilearn.problem_transform import LabelPowerset
from sklearn.metrics import f1_score,accuracy_score
lp_classifier = LabelPowerset(LogisticRegression())
lp_classifier.fit(X_train, y_train)
lp_predictions = lp_classifier.predict(x_test)
print("Accuracy = ",accuracy_score(y_test,lp_predictions))
print("F1 score = ",f1_score(y_test,lp_predictions, average="micro"))

alabrashJr

labels=["Fiyat","Geri Bildirim","Kullanım","Tavır ve davranış","Kalite"]
y_train=train[labels]
y_test=test[labels]

alabrashJr

from sklearn.feature_extraction.text import TfidfVectorizer
vectorizer = TfidfVectorizer(strip_accents='unicode', analyzer='word', ngram_range=(1,3), norm='l2', max_features = 10000)
vectorizer.fit(train_text)
vectorizer.fit(test_text)
X_train = vectorizer.transform(train_text)
x_test = vectorizer.transform(test_text)

alabrashJr

from sklearn.model_selection import train_test_split
train, test = train_test_split(data, random_state=42, test_size=0.30, shuffle=True)
train_text = train['TurkishStemmer'].values.astype('U')
test_text = test['TurkishStemmer'].values.astype('U')

alabrashJr

from nltk import word_tokenize
from nltk.tokenize import RegexpTokenizer
#!pip install TurkishStemmer
from TurkishStemmer import TurkishStemmer
stemmer = TurkishStemmer()
tokenizer = RegexpTokenizer(r'\w+')
punct_re=lambda x :" ".join(tokenizer.tokenize(x.lower()))

def stemmer_char(text,i):
    return " ".join([word[:i] for word in word_tokenize(text)])