Sentiment Analysis
2.92 billion people spend an average of 22.8 hours a month on the internet (Internet Usage Worldwide – Statistics & Facts), often sharing opinions via outlets such as social media, review sites, blogs and forums. This abundance of information could be collected and analyzed to provide insights such as consumer opinions, voters’ preferences, and incipient cultural shifts. Analyzing these data is referred to as sentiment analysis and can be performed by assigning positive and negative values to individual words or groups of words.
Sentiment analysis is currently in its early stages while developers try to overcome obstacles such as complex sentence structures and the connotation of words in different contexts. When these obstacles are overcome, this technology has limitless potential to strengthen the impact online consumer-created content has on brands, websites, social movements and even elections. Nonetheless, even humans only agree on the sentiment of text 79% of the time (Ogneva), so even the best sentiment analysis technology could not have 100% accuracy.
Sentiment analysis relates to our learning of MySQL and SAS Enterprise Miner in Data Analytics. We queried data to draw conclusions about its meaning. Similarly, sentiment analysis looks at writing, which is the data, and draws conclusions from it. For example, an analyst will be able to find an individual’s political views or opinions about certain products, just as analysts can use Workbench to query an individual’s address or purchases. Building on this framework, sentiment analysis also uses decision trees like the trees generated by SAS Enterprise Miner. Determining the sentiment of a post requires looking at several factors including sentence structure, context and conflicting sentiments. To overcome this challenge, a decision tree can be used to look at the specific word then surrounding words and context.
A potential use of sentiment analysis is determining public response to a product. Crawlers can search online and analyze the content on websites to determine what people think of a product. This analysis will enable companies to respond quicker and more accurately to consumer feedback. For example, a company could find out that consumers often complain about a product’s wiring or inaccurate marketing and alter its corporate strategy to improve customer satisfaction.
Works Cited
“Deeply Moving: Deep Learning for Sentiment Analysis.” Stanford. Stanford University, n.d. Web. 28 Mar. 2015. <http://nlp.stanford.edu/sentiment/>.
Ogneva, Maria. “How Companies Can Use Sentiment Analysis to Improve Their Business.” Mashable. Mashable, 19 Apr. 2010. Web. 28 Mar. 2015. <http://mashable.com/2010/04/19/sentiment-analysis/>.
“Statistics and Facts on Internet Usage Worldwide.” Statista.com. Statista, Inc., n.d. Web. 28 Mar. 2015. <http://www.statista.com/topics/1145/internet-usage-worldwide/>.