Content Intelligence at Booking.com

Important Disclaimer

In case you want to arrive physically to the event, you should have Tav Yarok and subscribe in the following link:
https://forms.gle/6umLBU9y7xpxXSCNA

Note that filling the form does not guarantee you will be able to enter the event.
Prior to the event we will send confirmation emails to those of you who signed first and can make it. The event will be recorded for those that won’t be able to attend in person.
--------------------

Excited to share with you our next physical face-to-face meetup at Booking.com Tel Aviv!

We’ll be hosting Booking.com CTO, Rob Francis, visiting us from Amsterdam, and 2 talks from Product and ML around the topic of Content Intelligence and enrichment.

The event will take place at the beautiful Booking.com Tel Aviv offices in Azrieli Town (near Midtown).
The event will be recorded for those that will not be able to attend in person.

About the talks:
Being a leader in the online travel industry, Booking.com has billions of images and other types of travel and user generated content that can be used to help travelers with decision making.
With recent years’ technology advancement and the introduction of deep learning technologies it is our mission to understand, enrich, and serve the content in a way which is useful for accommodation owners and travelers.

The evening will open with Booking CTO, Rob Francis, followed by an overview on the work done at Booking.com Tel Aviv and continue with Product and ML.

Agenda, Speakers & Talks:

17:30-18:00: Gathering, food, and drinks

18:00-18:15: Opening: “ML Products for the Travel Marketplace”, Noa Barbiro, Group Product Manager Machine Learning

18:15-18:30: Fireside chat with Rob Francis, Booking CTO

18:30-19:00: “Cracking the mystery behind ML platforms: How to build a Content ML platform in 3 steps”, Hadas Harush, Sr. Product Manager Machine Learning

In my talk, I’ll walk you through our product journey of building a content ML platform, a journey which has started from a high level definition and ended up being a fully functioning and scaled product which serves many teams within Bokking.com. I’ll go over 3 main steps towards achieving this goal and will share our challenges and dilemmas along the way.

Bio: Hadas is a Senior Product Manager Machine Learning at the ML center of Booking.com in Tel-Aviv. She has previously worked in global tech and eCommerce companies (ebay, HP and more) in different data and product management roles. At Booking.com she combines her 3 passions - product management, data science and the travel industry.
Hadas holds an M.Sc. in Information System Engineering with focus on Data Mining and BI and B.Sc in Industrial Engineering and Management.

19:00-19:30: “How much is an image worth? Predictive algorithm for image quality”, Karen Lastmann Assaraf, Sr. Machine Learning Scientist

On the word scale, we know an image is worth a thousand. But can we measure how much an image is worth in terms of quality?
Images play a huge role in understanding the value of a property.
Quality of photos, especially low quality, has an impact on the booking behavior and estimated price from the user perspective.
Being able to measure the quality of each image uploaded by our partners is a crucial capability. One can think of image quality in terms of aesthetic or technical attributes.
We will first explain why we decided to focus on the technical quality aspect, share more about the datasets used, the pretrained vision transformer architecture and the evaluation process.
Then we’ll share model predictions on Booking.com images and its potential impact all along the funnel: from main image selection optimization, to the property gallery subset selection, to the entire gallery ranking to giving our partners insight on the images they upload.

Bio: Karen is a Senior Machine Learning Scientist at the ML Center of Booking.com.
Her work focuses on computer vision models leading to business impact (image quality, image enhancement, image understanding).
Prior to Booking, Karen developed obstacle detection models in the autonomous car and autonomous train industries (Yandex, Railvision), and machine learning models for speech and NLP applications in a startup (Chorus.ai acquired by ZoomInfo).
Karen holds an M.Sc. and B.Sc in Applied Mathematics from TelecomParisTech and a business double degree with Ecole Polytechnique.