Quickly find out and respond to what people are saying about you and your organization.
This article compares and contrasts various cloud service utilites that provide Sentiment analysis of what’s being said about an individual, a brand name, or subject. Such tools provides a real-time way to guage how the strength of opinions – how positively or negatively – the public feels about specific brands, products, or services.
Sentiment analysis assesses the intent, tone, and emotion behind what the public posts.
IMPORTANT: Sentiment analysis is a difficult task because it involves human emotions. Research shows that humans will disagree about the sentiment of written text in about 20% of all cases. This means that even if the sentiment analyzer were a perfect tool, as a human being you would likely only agree with its conclusions about 80% of the time.
Sentiment analysis tools are powered by natural language processing (NLP) and machine learning (ML) to assign a sentiment score to each posting. Use of positive words would have a high score. Use of negative words would have a low score.
More sophisticated sentiment analysis tools are able to differentiate positive and negative feelings about different topics within a single sentence. Take this example: “I love their products but I’m not a fan of their recent ads.”
Most tools cannot differentiate sarcasm.
Some tools can differentiate between different sentiment in a sentence that refers to different points – classifying such statements as having neutral sentiment.
Most tools work only in American English language.
Most tools produce an overall sentiment score. So even though various passages within a sample of text may be particularly positive or negative, the sentiment score produced by the tool reflects all of the text taken as a whole.
Sentiment Analyzer at https://www.danielsoper.com/sentimentanalysis/default.aspx
This is a web page. Type or paste text into the box (replacing the example) and click the “Analyze Text!” button.
Sentiment analysis uses “computational linguistics” and “text mining” techniques to determine the sentiment or affective nature of the text being analyzed.
This tool was trained using the collection of more than 8,000 writing samples and transcripts of spoken conversations that appear in the American National Corpus (ANC). The ANC contains writing samples from a wide variety of genres and domains. So this sentiment analyzer is most accurate with text written in American English after 1990.
This tool computes a single sentiment score that reflects the overall sentiment, tone, or emotional feeling of your input text. Sentiment scores range from -100 to +100, where -100 indicates a very negative or serious tone and +100 indicates a very positive or enthusiastic tone.
NCSU Tweet Visualizer at https://www.csc2.ncsu.edu/faculty/healey/tweet_viz/tweet_app/
This tool generates a two-dimensional display of plesantness and active/subdued.
SentiStrength at http://sentistrength.wlv.ac.uk/ took too long to repond, so is considered dead.
Some tools have been trained on specific domains (e.g., business, religion, entertainment, politics, etc.).
Keyhole can track sentiment around each specific hashtag or keyword.
Keyhole also offers influencer tracking, social listening, and social publishing.
Others offer deep text mining features of enterprise-level solutions—which aren’t needed by the majority of marketers.
Keyhole pricing: Starts at $79 per month after a free trial.
https://keyhole.co/blog/best-sentiment-analysis-tools/
This is one of a series on macOS (Mac OSX):