Mitigating cold-start situations is a fundamental problem in almost any recommender system. In real-life, large-scale production systems, the challenge of optimizing the cold-start strategy is even greater.
We present an end-to-end framework for evaluating and comparing different cold-start strategies. By applying this framework in Outbrain’s recommender system, we were able to reduce our cold-start costs by half, while supporting both offline and online settings. Our framework solves the pain of benchmarking numerous cold-start techniques using surrogate accuracy metrics on offline datasets – coupled with an extensive, cost-controlled online A/B test.
In my talk, I’ll start with a short introduction to the cold-start challenge in recommender systems. Next, I will explain the motivation for a framework for cold-start techniques. I will then describe – step by step – how we used the framework to cut the size of our exploration data by more than 50%.
 

Moran Haham is an algorithms manager at Outbrain, with over 10 years of experience in the tech industry and data science. Moran leads recommendation system research, particularly click-through rate prediction on large scale in production. Moran is an international public speaker who believes that data-driven culture is what drives a company to perform at its best. She carries out the vision by using data and ML to enhance business success. Moran has a Master's degree in Information Systems from Ben Gurion University where her thesis focused on using active learning for rating elicitation in recommendation systems. Moran is alumni of successful startups such as SparkBeyond and Perion Networks where she solved complex problems including product recommendation widgets for e-commerce sites, demand forecasting, revenue forecasting, text classification. Moran has a proven track record of leading a team of data scientists delivering innovative ML models to fortune 100 companies from a wide range of industries including retail, finance, banking, and logistics.

What if you were able to know what are the most frequent topics our guests and partners are talking about? Get themed-based reviews? Understand the most important facilities for our guests? Have the most significant attributes and endorsements of a property or a destination? Detecting fraudulent messages? And many more..
In this talk, we will present how we build a multilingual multi-label topic classification model that supports zero-shot, with multiple architectures, using cross-encoder and bi-encoder.
We will show how fine-tuning BERT-like models on the tourism domain with a small dataset, can outperform other pre-trained models.

Moran is a manager of machine learning at booking.com, researching and developing computer vision and NLP models for the tourism domain. 
Moran is a Ph.D. candidate in information systems engineering at Ben Gurion University, researching NLP aspects in temporal graphs. 
Previously worked as a Data Science Team Leader at Diagnostic Robotics, building ML solutions for the medical domain and NLP algorithms to extract clinical entities from medical visit summaries. 

In recent years, factorization machine models (most notably FM and FFM) have dominated the realm of CTR prediction due to their ability to extract good user and item representations in sparse setups. While FFMs can produce more specific interactions between the different features, FMs are smaller models and are therefore the better choice for extremely sparse problems. There are, however, solutions that aim to get the best of both worlds. In this talk we will cover the basics of FM models, and present a hybrid variant which we used to better fit our production needs without loss of online performance. Finally, we’ll see how this solution can be integrated with alternative approaches from current literature.

Maoz is a Machine Learning group lead at Taboola, responsible for the machine learning and algorithmic aspects of “HP4U” - Taboola’s solution for personalizing publisher homepages, and “Taboola News” - Taboola’s news aggregation experience. Maoz holds an MSc in applied mathematics from Tel Aviv university and a BSc in mathematics and economics from Ben Gurion university.