Overview
The Semantic Analysis engine identifies and categorizes customer opinions to uncover meaningful insights. Each opinion consists of:
- Text: What the customer said, for example “The room was spotless.”
- Sentiment: The sentiment behind what the customer said: positive, negative, or neutral.
- Subcategory: What the opinion is about, in the example it will be “Room Cleanliness.”
The system assigns each opinion to a specific subcategory and clusters them into meaningful categories. These categories help aggregate and filter opinions for deeper analysis.
Benefits
- Sector-wide insights: Our Semantic Analysis is delivering actionable insights across multiple hospitality sectors such as Accommodation, Food & Beverage (F&B), and Points of Interest. This guarantees the right relevance of information for each specific Sector.
- Enhanced accuracy: AI-powered technology improves sentiment detection, offering better quality and increased coverage of customer opinions.
- Multi-language support: Analyze feedback in 23 languages, ensuring global applicability.
Categories, Subcategories and Departments
Subcategories
As mentioned above, the subcategories refer to what an opinion is about and can include:
- Physical spaces: Room, Bathroom, Bar.
- Tangible items: Meals (e.g., Breakfast, Lunch), TV, Furniture.
- Services: Room Service, Shuttle Service, Loyalty Program.
- Personnel: Cleaning Staff, Wellness Staff.
- Practices: Disinfection & Sanitization, Sustainable Sourcing.
- Characteristics: Cleanliness, Price, Size, Availability, Accessibility.
- Special cases: broad statements (e.g., “It was great.”) and other topics that don’t fit predefined categories.
Subcategories follow a MECE (Mutually Exclusive Collectively Exhaustive) principle. This means that opinions linked to subcategories are specific and don't overlap with each other.
Each subcategory consists of two components:
- Aspect: The first part of the subcategory name, referring to physical or tangible elements that can be evaluated.
- Quality: The second part of the subcategory name, describing the qualitative or intangible characteristics of the aspect.
For a detailed explanation of each subcategory, please see: Subcategories - Explanation and examples
Categories
Categories group related subcategories to enable filtering and aggregation of opinions by key areas.
You start your sentiment analysis always at a Category level, for example "Room", to later drill down into the detailed subcategories related to the Room, for example "Room - Cleanliness".
Keep in mind:
- Some subcategories belong to multiple categories, as they represent the same information viewed from different perspectives. For instance, the subcategory “Room – Cleanliness” is part of both the “Room” and “Cleanliness” categories.
- Categories can focus on either aspects (e.g., physical spaces or services) or qualities (e.g., cleanliness or size). Note that aspect categories already include all related quality subcategories.
This structure ensures flexibility and clarity in organizing and analyzing opinions.
For a detailed list of all categories and subcategories for each sector, please see: Categories and Subcategories mapping
Departments
Similar to categories, departments group related subcategories to enable filtering and aggregation of opinions by key areas. The main difference with categories, is that departments group the subcategories around a functional department and they can mix subcategories from multiple categories.
There are 5 defined departments:
- Maintenance
- Housekeeping
- Food and Beverage
- Front Office
- Wellness and Spa
If you want to understand which subcategories are mapped under each departments, please check the article: Departments and Subcategories mapping
Supported Languages
The Semantic Analysis engine is available in 23 languages, ensuring robust global functionality:
- Arabic, Chinese (Simplified & Traditional), Croatian, Czech, Danish, Dutch, English, Finnish, French, German, Hebrew, Indonesian, Italian, Japanese, Korean, Norwegian, Polish, Portuguese, Russian, Spanish, Swedish, Thai, and Turkish.