computer analysis can be very literal. Text is analyzed using a database that tells the computer whether certain words or phrases are positive, negative or neutral. Throw in slang terms, Internet-style speak and an evolving language, and computer text analysis isn’t 100 percent accurate.
Consider the phrase, “This is one bad hot sauce! Love it on a hamburger.” Automatic analysis may rate the message as neutral, because the words “bad” and “love” cancel each other out. A person would realize that in this context, the word bad is positive. While automatic sentiment analysis can be customized to recognize some phrases in context, the amount of customization and maintenance work may not be worth it when human analysis is more accurate.
People-powered processes offer added value for sentiment analysis functions. Rating a series of comments, words, phrases or sentences as positive or neutral is a simple task, which means you can add to it when humans are on the job. More complex sentiment analysis lets you categorize feedback by a things such as products or service, method of delivery or demographics. By mining user profiles or other online information, crowd workers can build customer profiles on top of performing sentiment analysis.
The information gathered from your website, customer comments or reviews can be customized almost endlessly when you use crowdsourcing. Find out more about crowd-powered sentiment analysis, as well as other data and categorization services, at CrowdSource.
Click to close