Skip to content

goutham-sp/scrappers

Folders and files

NameName
Last commit message
Last commit date

Latest commit

8a4a8e3 · Apr 26, 2018

History

6 Commits
Apr 26, 2018
Apr 26, 2018
Apr 26, 2018
Apr 26, 2018
Apr 26, 2018
Apr 26, 2018
Apr 26, 2018
Apr 26, 2018
Apr 26, 2018
Apr 26, 2018
Apr 26, 2018

Repository files navigation

scrappers

word2vec scrapers:

w2v_scraper.py: Scrapes data off sense2vec output trained with reddit dataset
	Scrapes data off https://explosion.ai/demos/sense2vec.
	The model that is used is trained is up-to-date. Scraping off of this is more accurate than the library.

w2v_scraper_gnews.py: Scrapes data off sense2vec output trained with google news dataset
	Scrapes data off https://rare-technologies.com/word2vec-tutorial/
	The model used is trained with google news dataset.

altpick.py : Scrapes data of photographers of specific location
	https://altpick.com/ is the source website.

asmp.py : Scrapes data of photographers of specific location
	hhtps://asmp.org/ is the source website
	
wonderfulmachine.py : Scrapes data of photographers of specific location
	https://wonderfulmachines.com/ is the source website

production_paradise.py : Scrapes data of photographers of specific location
	https://www.productionparadise.com/ is the source website
	
company_page.py : Scrapes all useful information from a company's website
	Any page with important info can be scraped

producthunt.py : Scrapes data about a company from https://producthunt.com

whois.py : Scrapes data from whois website for specified company website

wiki.py : Looks up about a company on wikipedia

Releases

No releases published

Packages

No packages published

Languages