class: center, middle, inverse, title-slide .title[ # Sentiment Analysis Applications ] .author[ ### David Garcia <br><br> <em>ETH Zurich</em> ] .date[ ### Social Data Science ] --- layout: true <div class="my-footer"><span>David Garcia - Social Data Science - ETH Zurich</span></div> --- # Applications of sentiment analysis </br> ## 1. Example applications to art and design ## 2. Applications in the digital humanities ## 3. Applications in business and finance --- # London eye .center[] - London Eye showing sentiment in Tweets during the 2012 Olympics - The output of SentiStrength was converted to the color over the ferris wheel --- # Automated album covers .center[] - Visualization of emotions in the titles of songs of an album - Word classification combining ANEW and SentiStrength --- # Digital humanities: Music lyrics .center[] - Application of ANEW lexicon to lyrics of songs since the 1960's - Downward trend replicated in several later articles --- # Syuzhet: plot sentiment .center[] - Application of MPQA lexicon to the text of novels - Used to identify the six patterns of plots theorized by Kurt Vonnegut --- # Google books misery .center[] - Literary misery in Google Books: LIWC NA score (Germany example) - Literary misery is correlated with economic misery of the previous decade --- # Business: Sentiment about products .center[] - Netbase Brand passion indices using Twitter, Facebook, product reviews... - Vaguely documented sentiment analysis: rule-based method with lexicon --- # Twitter mood and the stock market .center[] - MPQA lexicon, also called OpinionFinder, applied to "I feel" tweets + adaptation of POMS (Profile of Mood States) - Predicting movements of the Dow Jones Industrial Average (DJIA) using sentiment aggregates from tweets (Bollen et al, 2011) --- # Trump2Cash .pull-left[ - Google NLP API to classify sentiment about companies in Trump's tweets and trade - **What do you think was its return in early 2017?** ] .pull-right[] ---  --- # Summary of day 3 - **The Semantic Differential** - Quantifying meanings of symbols through adjective ratings - Dimensionality reduction shows three dimensions: Evaluation, Potency, and Activation - **Supervised Sentiment Analysis** - Using labelled texts to train a model - Text representation and model selection are important - Evaluation metrics: Precision, Recall, `\(F_1\)` - **Applications of Sentiment Analysis** - Applications to art, digital humanities, and business - Always think whose sentiment you are measuring!