Google Code-in

Given a bunch of captions extracted from the newscast, identify the sentiment associated with each caption or sentence. Sentiment analysis basically requires identification of polarity of a sentence (often as simple as positive, negative and neutral). This task specially helps in analysing the sentiments of the news broadcast.

You basically needs to train a classifier that will classify any input test sentence into any of the three classes i.e neutral, positive or negative. You need to use news corpus(data) for training the classifier. You can use any external libraries for text processing. Libraries like nltk and gensim will be helpful. Have a look at this article: https://datadrivenjournalism.net/featured_projects/trump_and_the_media_a_sentiment_analysis_of_news_articles