Feed Inazuma Eleven

I was occupied with office work. and read this paper Day 6/30: "Field-aware Calibration" (Pan et al., WWW 2020) A recommendation model with a great AUC score can still make wildly unreliable predictions over specific user or item groups #30Days30Papers #RecSys #MachineLearning

Built CineMatch 🎬 an explainable movie recommender. Pick 5 films → get recs, each with a "Why this?" breakdown of the real scoring. ALS + kNN + MiniLM + MMR · FastAPI + React Live: t.co/4QeiVjNL1Y Code: t.co/cIb3CNreNp #MachineLearning #RecSys #Python

I didn’t make it to @ACMRecSys this year, but very happy to receive the 𝐎𝐮𝐭𝐬𝐭𝐚𝐧𝐝𝐢𝐧𝐠 𝐑𝐞𝐯𝐢𝐞𝐰𝐞𝐫 𝐀𝐰𝐚𝐫𝐝! Grateful to have contributed to the community! And a very 𝐇𝐚𝐩𝐩𝐲 20𝐭𝐡 𝐀𝐧𝐧𝐢𝐯𝐞𝐫𝐬𝐚𝐫𝐲 to @ACMRecSys #Recsys t.co/6gc3elefqB

Are attention/transformer models a dead end for sequential recommendations? It seems so. Sometimes they are on par with other approaches but in most cases they fall behind. Even a simple Markovian beats most & a stateless model beats half. #RecSys Details: t.co/hjIKIrYgPA

20 years after the Netflix Prize, we tried again. Modern hardware. Deep learning. AI-assisted coding. Published solutions. First-hand competition experience. Surely the 10% target would be easier today? We got to 9.98%. Still hard. 🧵 #RecSys #MachineLearning @ACMRecSys t.co/1JYMr8qAYL

Well, it seems #RecSys is once again in a position to adapt to another round of change. t.co/1dXY1NP5oz

Yesterday: notebook → production Today: made it real • SSOT architecture • Fixed data leakage properly • Model optimization + robustness • Serving layer ready Not flashy. Necessary. This is what “notebook to production” actually looks like. More to do. #MLOps #RecSys #AI

This reminds me of some of the research in #RecSys

A Comprehensive Tutorial on #RecSys Engines and Implementation! #BigData #Analytics #DataScience #AI #MachineLearning #IoT #IIoT #Python #RStats #TensorFlow #Java #JavaScript #ReactJS #CloudComputing #Serverless #DataScientist #Linux #Programming #Coding #100DaysofCode t.co/s2Wxk9cPdp

Delighted to see my #RecSys paper featured in Advances in Computing by ACM! When Science Goes Agentic t.co/dmwnCgGVBU

So glad to learn that we both share the very same vision for the future of #RecSys! Both position papers focus on how AI agents change who we are recommending to. The Agentic Web is here, and RecSys needs a paradigm shift! t.co/oSNITEEA1D

I'll be giving an invited talk on our work on Multimodal LLMs with Xiaohongshu at the DaQuaMRec workshop at #RecSys2026 on September 28 in Minneapolis, MN! 🇺🇸 Many thanks to the organizers. Looking forward to it! @ACMRecSys @sjtu1896 @xiaohongshu @DaQuaMRec #RecSys #LLM #MLLM #MultimodalLLM #Recommendation

Solving-Cold-Start Conundrum in #RecSys! #BigData #Analytics #DataScience #AI #MachineLearning #IoT #IIoT #PyTorch #Python #RStats #TensorFlow #Java #ReactJS #GoLang #CloudComputing #Serverless #DataScientist #Linux #Programming #Coding #100DaysofCode t.co/d2EPhmrlta t.co/tncBrD5A1W

While #SIGIR2026 is underway, I'm happy to see that our short paper, published six years ago, has now reached 100 citations on Google Scholar. Looking at recent #RecSys research, I see a healthy shift: beyond proposing new algorithms, the community is increasingly re-examining the foundations—baselines, reproducibility, datasets, evaluation, the academia–industry gap, and the assumptions behind our benchmarks. The next challenge may be even bigger. As AI agents increasingly act on behalf of users, we need to rethink not just recommender algorithms, but the entire RecSys ecosystem. If AI agents generate a significant share of internet traffic, recommendation, search, and their business models will all need to be reimagined. That transformation is already underway with the rise of agentic search (e.g., @octenai) in the search domain, and RecSys is likely to follow. In the future, recommender systems may not only be built by AI agents but increasingly consumed by AI agents acting on behalf of users.

Dear #recsys #recsys2026 community, my team at Penguin Random House 📚 (one of the biggest book platforms) is hiring a Staff ML Scientist – Personalization. If interested, plz reach out to me or the hiring manager K.T. Chang (t.co/uAfgBLnjTU) job:t.co/U5fLiGHop2

TorchRec (Beta Release) for #RecSys. #BigData #Analytics #DataScience #AI #MachineLearning #IoT #IIoT #PyTorch #Python #RStats #TensorFlow #Java #JavaScript #ReactJS #CloudComputing #Serverless #DataScientist #Linux #Programming #Coding #100DaysofCode t.co/6jcydUySS4 t.co/tXUF5SQPXZ

Building #RecSys WorkShop! #BigData #Analytics #DataScience #IoT #IIoT #PyTorch #Python #RStats #TensorFlow #Java #JavaScript #ReactJS #GoLang #CloudComputing #Serverless #DataScientist #Linux #Books #Programming #Coding #100DaysofCode t.co/DwzULKdPc4 t.co/qO2xd7T7IA

🎉 Two papers at #SIGIR2026! 🐱 CATS: Clustering + Thompson Sampling for smarter negative mining in RecSys. +20% NDCG@10. ✂️ Load-sensitive Selective Pruning: dynamic embedding dimensionality for dense retrieval under load. +400% queries served. See you in Melbourne! #IR #RecSys

Using Reviews to Create a #RecSys That Works. #BigData #Analytics #DataScience #AI #MachineLearning #NLProc #IoT #IIoT #Python #RStats #TensorFlow #JavaScript #ReactJS #CloudComputing #Serverless #DataScientist #Linux #Programming #Coding #100DaysofCode t.co/chceDmHs5l t.co/08jrduvg24

It's hard. The same goes to #RecSys t.co/vHp1585QHe

Computational Linguistics Journal@CompLingJournal

What % of the NLP papers measure their impact in the real world? This paper proposes an "impact evaluation" of NLP models or systems for real-world usage, changing the research culture of NLP to focus more on real-world impact and less on SOTA-chasing: https://t.co/3Ti5vUgdgs https://t.co/sisisF9WyY