The sentiment score is a measure of a person’s affinity of a word they mentioned. It is scaled from -100% to 100%. On the low end, -100% is extreme dislike for the word, 0 is factual/neutral and 100% is high support.
Our sentiment analysis considers weights of the individual words, the positioning of the words and the typical interpretation of the phrasing of the overall sentence, and depending on the situation, the context of the phrase (for example, well received comments on press releases of scandal are often satirical therefore such comments are interpreted with a slightly different algorithm)
The sentiment analysis data set is continuously trained by real people manually rating common phrases (similar to how captchas help digitize books)
We sample random users who mention the word you’re interested in across public comments on Twitter, Facebook, Reddit, Yahoo, Bloomberg and other various news sites. Each time you search, a new sample is taken across the internet.
85% of the time, the algorithm is correctly able to determine whether someone was generally speaking positively, negatively or neutrally about a term.
The line graph is the average sentiment score at the given date and time
- The average sentiment score gives a good indication of how well received something is by the general population.
- Simply put, you’ll be able to know how people feel about a brand, product or ideal without the need to lose copious money and time on experimentation
- Dips and spikes in the graph can be examined to show the cause of such sudden changes in public perception at a rate far greater than traditional surveys
The list allows you to get a deeply detailed look at some of the phrases sampled to obtain the sentiment score. You can hover over a username to see the sentence written by the user.
Sometimes, you may even be able to directly message the account and either ask a question, request a donation, make a sale or do whatever is relevant to your organization, often with a much higher success rate since the person is being contacted from a psychologically relevant angle versus an impersonal measure such as race, age or gender.
Sometimes a term either has an abundance of information about it online (in which case it can take a couple minutes to populate everything) or it does not have enough data. In the former case, the terms process in the background and you’ll get a message when it’s ready. If there isn’t enough data, then you’ll be notified of that too