Case Studies

‘Rex’: Medium’s Recommender-System as a Microservice [Miles Hinson]

Miles Hinson writes about his experience in developing ‘Rex’, the recommender system for Medium, as a Microservice. He discusses design choices and presents a few numbers such as that 95% of all recommendations are created within less than 1 second. However, the biggest issue was language choice. Much of Medium runs within a Node.js monolith, […]

Posted in Case Studies | Tagged , , , | Leave a comment

Against All Odds: Netflix, From Mail-Order DVD Rentals to Streaming Dominance [Shawn Knight @ TechSpot]

Shown Knight from Techspot writes about the history of Netflix and how Netflix evolved from a mail-order DVD rental to the world’s largest streaming service. Home video rentals were already a $16 billion industry when Reed Hastings and Marc Randolph decided to get involved in the summer of 1997. Hastings, who holds degrees in mathematics […]

Posted in Case Studies, History | Tagged , , | Leave a comment

What Twitter learned from the Recsys 2020 Challenge [Michael Bronstein et al.]

‘What Twitter learned from the Recsys 2020 Challenge’ is a blog post authored by Michael Bronstein, Luca Belli, Apoorv Sharma, Yuanpu Xie, Ying Xiao, Dan Shiebler, Max Hansmire, and Wenzhe Shi; published on TowardsDataScience. The authors describe the RecSys 2020 Challenge that was performed on Twitter data. [We] describe the dataset and the three winning […]

Posted in Case Studies | Tagged , , , , , , , , , , , , | Leave a comment

Spotify considers the interests of artists when recommending music to users

Spotify announced a new feature that allows artists and music labels to influence how their music is recommended to the users of Spotify. In this new experiment, artists and labels can identify music that’s a priority for them, and our system will add that signal to the algorithm that determines personalized listening sessions. This allows […]

Posted in Case Studies, Example Applications | Tagged , , | Leave a comment

Case study how simple recommendation techniques improve user experience in e-Commerce [Emilio Carrión @ MercadonaTech]

Emilio Carrión (MercadonaTech) wrote a blog post about how simple recommendations to substitute ‘ghost’ products improve conversion rate and user experience. He defines ‘ghost’ products as items that are available at the time a user adds them to the shopping cart, but are not available any more when the user proceeds to check-out. It seems […]

Posted in Case Studies | Tagged , , | Leave a comment

Facebook documents its recommender system — a little bit. [Sarah Perez @ Techcrunch]

Facebook has not been very open about how its recommender system works exactly. This may be about to change though. Today, Techcrunch announced that Facebook “has, for the first time, made public how its content recommendation guidelines work”. However, Techcrunch also points out that there is little technical detail on how recommendations work. Algorithmic recommendation […]

Posted in Case Studies, Documentation | Tagged , , , | Leave a comment

The Architecture Used at LinkedIn to Improve Feature Management in Machine Learning Models [Jesus Rodriguez@Medium]

The scale of the machine learning problems that an organization like LinkedIn deals with results incomprehensible for data scientists. Building an maintaining a single, effective machine learning models is hard enough so imagine coordinating the execution of thousands of machine learning programs to achieve a cohesive experience. Feature engineering is one of the key element […]

Posted in Case Studies | Tagged , , , | Leave a comment

How Do Product Attributes and Reviews Moderate the Impact of Recommender Systems Through Purchase Stages? [Dokyun Lee & Kartik Hosanagar]

An interesting study on the impact of recommender systems in e-Commerce. While, on average, views, conversion|views, and final conversion rates varied by 15.3%, 21.6%, and 7.5%, respectively, the numbers strongly depended on the product attributes. The study is based on an A/B test (recommendations vs. no recommendations) with 184,375 users. We find that the lift […]

Posted in Case Studies | Tagged , , | Leave a comment

Google medical researchers humbled when AI screening tool falls short in real-life testing [Devin Coldeway @TechCrunch]

Google achieved great ‘theoretical accuracy’ on the detection of diabetic retinopathy. However, in the real-world their approach didn’t perform so well. While this is not directly related to recommender systems, it illustrates a problem that also exists in recommender-system research: Algorithms may perform excellently on some offline datasets, yet they (sometimes) fail to perform well […]

Posted in Case Studies | Tagged , , , | Leave a comment

Why Recommendations On Netflix, Amazon, Or WeChat Could Be More Influential Than You Think [Forbes]

“When’s the last time you saw a movie in a cinema? How about the last time you watched a movie on Netflix? If you’re like most consumers, you’ve done the second thing much more often. But how much power do platforms like Netflix, Amazon, or WeChat, have on consumers and your potential customers? If you […]

Posted in Case Studies | Tagged , , | Leave a comment