[Week 1 — Hello Restaurant]

This is our first article. Let us to talk about that what will we do/work in this project.

The sentiment analysis is one of work area in machine learning and there are many related researches associated text/documents classification, spam filtering. Our project focuses the review of the restaurants. The main point of our project is collection our dataset in Turkish language and classify the reviews as accurate as possible in different categories using different approach such as Naive Bayes, Neural Networks, Support Vector Machines and etc.

Collection Dataset

First of all, our aim is to collect dataset and in this step, we have two different ways: collect manuel(parse HTML with Python) or use API(the interfaces provided by companies).

from lxml import html

import requests

page = request.get(<restaurant URL>)

tree = html.fromstring(page.content)

comment = tree.xpath(<comment tag>)

We think about yemeksepeti.com as company which is the biggest union food companies and has large database in terms of comments/reviews. They also provide several APIs (http://messaging.yemeksepeti.com/messagingwebservice/integration.asmx) for developers/users. We contacted the yemeksepeti.com to utilize this APIs and we have been waiting respond. So they will provide us username and password to use their library.

After the collection data, we think to do simple language processing which is like auto-correct at next step.

Stay tuned for news about our project.