The sentiment score is indexed on a normalized minus five (-5) to plus five (+5) scale. With this index, business users can roll up scores to preset categories, regions, or products, etc. to see sentiment trends across a brand, product, and/or company level.
Tune for Precision. The process of indexing the relative negativity (or positivity) is scored at a word, phrase, and linguistic construct level. This enables the system to identify complex linguistic constructs like negation, capitalization of words and other visual indicators of tone. Our sentiment analysis interface enables your analysts to further refine our out-of-the-box models to:
Modify a word's sentiment rating. Context can affect where words should appear along the negative or positive spectrum. The word thin for example, is positive if you are discussing your thin cell phone (unless, of course, it’s too thin—that would be bad). But if you say your hotel had “thin walls, thin blankets, and thin towels,” these are all negatives.
Handle negation. Clarabridge automatically inverts the sentiment for negators. This ensures that "not happy" is marked with negative sentiment rather than positive sentiment even though the word happy appears.
Use exception rules to specify context. Clarabridge has built-in exception rules for advanced tuning. For example, "would be <positive>" is a negative; "supposed to be a wonderful experience" should be negative rather than treating "wonderful" as positive. The word "friendly" is positive but the phrase "too friendly" is negative.