6 big brands who are using feature extraction in the wild

Unstructured data is really coming to the fore of people's minds," and the examples above all point that there is much to be done with unstructured data

Nick Millman from Accenture

1. AirBnb

AirBnb has decided to use feature tags derived from user reviews to aid visitors in sorting the listings by preference.

Tags such as Near public transit and Cozy help visitors to understand more about each listing at a single glance.

2. TripAdvisor

Trip advisor is one of the first companies to adopt feature extraction. The way that they have tackled the challenge is to show raw feature phrases. Examples of these are engineering marvel great drive and beautiful bridge

Once the tag is clicked the reviews are updated to reflect the matched opinion phrase. In this case engineering marvel has been selected.

Due to mobiles devices having much smaller screen size, this feature can really provide value by filtering the reviews that the user might only care about, whilst hiding the rest.

In this case garlic naan is what I am interested in...ummm.

3. Yelp

Unlike other companies Yelp has decided to display whole opinion snippets. You can find them directly under the restaurant's information.

However the functionality is still very similar to the rest of the competition, once a feature is clicked, the reviews adjust to show only the ones that mention that particular dish. In this case Pistachio Blackberry Croissant was clicked

4. Booking.com

Booking.com's implementation is something in between TripAdvisor and AirBnb. Visually it looks very similar to TripAdvisor but upon further investigation we find some key differences.

The tags are actually categories for themes that are mentioned in the reviews. So when you click on Ambience, the reviews that are displayed might not have the exact word "ambience" present but they will have something to do with the theme of ambience.

5. GlassDoor

GlassDoor is one of the first companies that operates in recruitment to implement feature highlights. The reviews are handled in a very similar fashion to Yelp.

Once the smart people feature is clicked the reviews filter to show only the ones that match the mention of "smart people"

6. Gumtree

Gumtree doesn't have very much opinionated content, but what they have is an abundance of product descriptions. These product descriptions are rich in features, so why not use them to improve search?

Once you search for "One Bedroom" complimentary tags like one bedroom flat and double room are displayed. This is great because the amount of effort the user has to exert to browse the site is drastically reduced.

Again the importance of this feature is massively increased on the mobile platform because there is less space to show the information that is relevant to the user.

Due to recent developments in machine learning, only now, are we able to unlock so much value from textual data. At SentiSum we help large enterprises in finding key insights from unstructured data. We are experts in the field of feature extraction. Get in contact when you are ready to take your business to the next level.