{"id":1921641,"date":"2014-02-18T18:08:41","date_gmt":"2014-02-18T12:38:41","guid":{"rendered":"http:\/\/stockviz.biz\/index.php\/?p=1921641"},"modified":"2014-02-18T18:40:20","modified_gmt":"2014-02-18T13:10:20","slug":"news-reading-sentiment-analysis-app","status":"publish","type":"post","link":"https:\/\/stockviz.biz\/index.php\/2014\/02\/18\/news-reading-sentiment-analysis-app\/","title":{"rendered":"A News Reading and Sentiment Analysis App"},"content":{"rendered":"

We previously discussed<\/a> how you can pull news items for specific topics\/stocks using the StockViz News API<\/a>. But all that the script did was show you a link to the news item. But who has the patience to click on a link, go to the website only to get annoyed by the ads? Wouldn’t it be nice if you could just read the news then and there?<\/p>\n

Goose<\/h3>\n

Python Goose<\/a> is an open-source article extractor written in python. It takes any news article and extracts the main body of the article and also all meta data. By adding a few lines of code, you can convert the previous script<\/a> into a news reader of sorts that displays just the article, minus the cruft.<\/p>\n

4 lines of code:<\/strong><\/p>\n

\n
\n\n\n\n\n\n\n
import the library:<\/small><\/td>\nfrom goose import Goose<\/small><\/td>\n<\/tr>\n
initialize goose:<\/small><\/td>\ng = Goose()<\/small><\/td>\n<\/tr>\n
extract the article:<\/small><\/td>\narticle = g.extract(url=r[‘SOURCE’])<\/small><\/td>\n<\/tr>\n
clean the article:<\/small><\/td>\ncleaned = article.cleaned_text.encode(‘ascii’, ‘ignore’)<\/small><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<\/div>\n

You can see the whole script here<\/a>.<\/p>\n

Sentiment<\/h3>\n

One of the advantages of Mashape is that there are number of other APIs that you can access from within your account. Most of these have a “freemium” plan, just like our API. One of these is the Free Natural Language Processing Service<\/a>. <\/p>\n

The NLP Service takes a URL and returns the sentiment (positive, negative, neutral) along with a sentiment score.<\/p>\n

2 lines of code:<\/strong><\/p>\n

\n
\n\n\n\n\n
make the call:<\/small><\/td>\nresponse2 = unirest.get(…)<\/small><\/td>\n<\/tr>\n
read the sentiment:<\/small><\/td>\nsentiment = response2.body[“sentiment-text”]<\/small><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<\/div>\n

You can see the whole script here<\/a>.<\/p>\n

Now you can tailor the what, when and how of the news you want to track. You could bundle all this up with a fancy user interface. Or pipe this through a messaging API (like SendHub<\/a>?) Chief ingredient: your imagination!<\/p>\n

Previously<\/h6>\n