What are not usable comments ?
Definition and general principles
The comments sent to the platform will be analysed through the solution in order to get:
- Emotion metrics like the detection of the main (primary) emotion, emotional intensity level, speech engagement level etc.
- Categories (depending of the data, the industry and the use cases)
and to propose you to get directly to the most actionable insights.
2 steps to qualify comments as "Not usable":
Some comments are eliminated from the results: they are called " not-usable comments". In average, less than 5% of the comments are not usable.
They are detected during 2 steps:
- At the data integration
In any data that haven't be checked manually, some noises may be present: the solution automatically eliminate the noise present into your data, and exclude it from your results.
Ie. the very short comments (less than 3 char.) are automatically excluded from the data.
- When the emotional analysis is performed
In addition to this first process, the remaining comments will be analysed and another algorithm will be launched in order to maximise the reliability rate of the emotional results.
Those 5% (up to 10%) excluded comments may contain more than 70% of the emotional mistakes.
This means that any comment will be classified as not usable:
- when the emotional markers are not clear enough, (Delta to define a main emotion)
- when the expressed emotions seem to be contradictory (Opposition into the emotional markers)
- when within a sentence no word is recognised using our dictionary (or even close to existing words) (Recognition performance rate)
This enables the solution to usually reach very high reliability rate (over 90% at the main emotion level for known industries and idioma).
Related Articles
How do you calculate the number of mentions in the Benchmark ?
General principles In the benchmark, you can see the number of mentions under every title. The mentions are the number of comments or Verbatim that is corresponding to the crossroad between filters and selection criteria (horizontal x vertical) you ...How are selected the elements of the card "Success points & points to improve" ?
The context of the card The card Success points & points to improve is situated into the module °Global > Global results. The card is closed to Tops & Flops comments card and the Word cloud card. The card is also proposed on the emotional mapping ...I find some symbols like \_ or _/ in the comments: why ?
Imagine that in a survey, your question had a negative (like in the example below) connotation. Imagine that the response of the customer is: Price and coverage. In this case, the emotion expressed was negative, but no markers are coming from ...If over a period my employee comments get a 23°C (73.4°F) emotional index, does that means that they are mainly in the surprise mood?
F A L S E . This is because of the average principle. When a class average score reaches 10, that doesn't mean that the whole class got a score of 10. If that is really the emotions that defines the emotional index, once the average have been ...What shall I do if new fields emerged in some filter columns of my data file(s) ?
General Principle When in a filter column of your data file, a new field is arising, the corresponding entry will be immediately and automatically accessible since the data processing. Therefore, if you add a new store or a new hotel in your filter ...