View Code? Open in Web Editor NEW

Bachelor Thesis: Classsification of Advertisements by means of Supervised Learning Methods

Python 55.43% C++ 0.25% C 1.13% Objective-C 0.04% XSLT 0.21% Roff 0.02% HTML 0.01% Shell 0.01% GAP 0.02% PowerShell 0.01% Batchfile 0.01% Fortran 0.01% Smarty 0.01% Jupyter Notebook 42.87%

diploma's Introduction

Diploma

My bachelor work in Hochschule Merseburg written in Python, using Native Language Processing of ML

My bachelor thesis is: Classification of advertisements by means of supervised learning methods

Work process:

Learn about NLP
Scrap data
Try NLTK / spacy on datasets
Learn more about hclustering algorithms / Neural networks / Other NLP methods like Topic Modelling, W2W and so on
Code the Diploma
Write a Diploma itself = Thesis

My bachelor has two major branches:

Data
- Scrapping data from web using scapy, google useragent or proxies. I used to scrap amazon with proxie, but because of lagging and switching off decided to use useragent and time.sleep()
ML
- Code implemenation

Commits

One of the 2 branches above: subproject: message. Not including README.md.

Example:

Data: amazon: added new spider

README.md: update

Data comes from these websites:

obszone
- had problems downloading american products for sale, so had to use a litle trick with url
geebo
adlandpro
pennysaverusa
hoobly
oodle
gumtree
letgo
salespider
ebay
amazon

Amazon data issues:

When entering departments on amazon you can scrap either 400 pages of common products of said department, or go into Feature Categories and scrap precise products.
For instance: 400 pages of automotive department OR Car care, car electronics and so on.

Recommend Projects