Sentiment analysis is the process of classifying text by identifying subjectivities expressed in it. For example, text can be classified as positive, neutral, or negative, and often expressed with a score to signify strength of the sentiment.
Examples of sentiment analysis.
Sentiment Analysis Applications
Sentiment analysis is used in almost all industries for applications such as:
- Identifying pain points and gaps for better product/process design using sentiment scores derived from customer surveys and social media
- Building an asset selection model for trading with sentiment scores of financial reports and news articles
Sentiment Analysis Techniques
Sentiment analysis uses text analytics, which combines natural language processing with machine and deep learning algorithms for building classification models and estimating sentiment scores. The two most common approaches for sentiment analysis are:
- Use prebuilt dictionary: You can start with an existing dictionary that categorizes different words in different polarities (such as positive/negative), emotions (such as angry/sad/dissatisfied), or numeric scale. You can then build a sentiment analysis model using the dictionary, predict sentiment in individual words in your text, and finally combine individual scores into an overall sentiment score for the text.
- Use prelabeled documents: You can also build a sentiment analysis model using deep learning techniques from a prelabeled dataset that categorizes each document (review, tweet, or other pieces of text) into different sentiments. You can then use the model to predict sentiment in a new document.
To learn more about importing, exploring, visualizing, and building models with text data including sentiment analysis, seeText Analytics Toolbox™.
