+31 20 – 752 03 16


16 sep 2022

Olery Sentiment Analysis Technology has gone through a fundamental rebuild and Sentiment V2 (Fusion of Data) is our latest and most advanced Sentiment Analysis Technology. Sentiment V2 analyses millions of reviews in multiple native languages. It detects the positive and negative sentiment towards specific Hospitality related Ratings, for example - Room and Personnel. Sentiment V2, then converts the captured sentiment into an easy to understand numerical score out of 10. The Ratings are lined up with the Numeral Rating system, Cleanliness, Facilities etc. This allows the Fusion of Sentiment and Numerical data for the first time within the Hospitality Industry. Capturing deeper and more translatable information.

Sentiment Analysis Technology has traditionally been based upon identifying positive and negative words within a written text or in the Hospitality Sector, Review. It then assigns a positive percentage to the whole review. This provides a very basic understanding of the sentiment as a whole, but fails to capture anything deeper and meaningful for the Hospitality Industry.

The next evolution was to take it a step further in identifying keywords within a Topic and provide a percentage of positivity on those keywords for that Topic. This required two separate lexicons of keywords and a more sophisticated detection technology, that is based per language. You have to map out the language structure to correctly identify the correct polarity word for the Keyword linked to the Topic. Within the Hospitality Industry, this is typically done only for English and other Languages are translated into English first. Also, it only covers Accommodations. The percentage system of scoring is also difficult if not impossible to link with numerical data. Olery Sentiment V1 was an expanded version of this with 6 Language structures mapped out instead of just English.

Olery Sentiment V2

Olery has taken Sentiment Analysis to the next level with V2 (Fusion of Data). Because we have gone Vertical, we are able to design a much more translatable scoring system based upon a more precise context based Lexicon (keywords) per rating. The Vertical Approach allows us to build more relatable and specific Keywords per rating. For example, Attractions Vertical, Facilities Rating, Attractions Sub-Rating, can now be extracted from the written reviews.

A variation functionality added to our Lexicons technology, allows us to navigate the often chaotic world of Online Reviews. From Typos caused by Fat Fingers or auto correct gone wrong, the variation allows us to capture these correctly to a much bigger degree than previously. An example of this is in English, the word Personal is often mis-used to mean Personnel when talking about staff in reviews. Variation allows us to link Personal to Personnel and capture the sentiment when in previous sentiment versions, this would have been missed.

Expanded Language Detection and Structures, allows us to grow our native language sentiment Analysis capabilities to currently 14 languages and expanding. Sentiment V2 (Fusion of Data) is the only Sentiment Analysis technology within the Hospitality Industry that covers this amount of languages natively.

Fusion of the Sentiment Scores and Numerical Scores is a Sentiment First. By converting the positive and negative sentiment into an industry established scoring system, we create more usable and actionable insights than ever before. Sentiment can now be extracted from Travel Composition, Nationality and much more.

A more structured approach that Sentiment V2 (Fusion of Data) has, allows for customers to play freely and easily with their own created Ratings. You build the keywords per rating and utilise our lexicon of Polarity words to extract sentiment that is related to your needs.

The Sentiment V2 (Fusion of Data) is not just an update of the current technology, it is a complete rebuild from scratch to enable the platform to run “stand alone”. We are now capable of offering Sentiment as a Service, enabling you to analyse whatever data sets you like to be analysed (eg guest communication via emails, whatsapp, chats etc).

Other blogs