Skip to content

Latest commit

 

History

History
36 lines (25 loc) · 1.86 KB

README.md

File metadata and controls

36 lines (25 loc) · 1.86 KB
Raw

NLDB'16 Sentiment Analysis

This is the demo code for the Sentment Analysis talk at the 2016 NLDB Conference in Salford

Requirements

All scripts are written in python, therefore a functioning python 2.7 environment is necessary. Dependencies are loaded via pip. If you do not have pip installed, this guide will provide detailed information.

Following packages are required:

  • numpy 1.9.2
  • matplotlib 1.4.3
  • nltk 3.1
  • scikit-learn 0.16.1
  • scipy 0.16.0b2
  • tqdm 3.7.1
  • facebook-sdk 0.4.0
  • requests 2.7.0

To make it easier for you, we have provided a requirements.txt which can be installed by moving into the checked out directory and running pip install -r /path/to/requirements.txt.

Analyze Facebook Pages

To analyze Facebook pages, run python classify.py. As an example to pull all posts from the Austrian railway operator ÖBB and filter them by the keywords flüchtling, krise, asyl from 1st of September until 30th of November following command is used: python classify.py unsereOEBB --access_token ABC...98AE --keywords flüchtling krise asyl --date_start 2015-09-01 --date_end 2015-11-30

Arguments

--access_token REQUIRED The access token is mandatory to make requests against the Facebook-API. A temporary access-token can be obtained here: https://developers.facebook.com/tools/explorer/

page REQUIRED Facebook page id to import

--date_start The starting date to import posts from the page - format YYYY-MM-DD.

--date_end The ending date to import posts from the page - format YYYY-MM-DD

--keywords List of keywords separated by space. Only posts containing these keywords will be classified and plotted.