Media
HypeRS: Building a Hypergraph-driven ensemble Recommender System
Gharahighehi, Alireza, Vens, Celine, Pliakos, Konstantinos
Recommender systems are designed to predict user preferences over collections of items. These systems process users' previous interactions to decide which items should be ranked higher to satisfy their desires. An ensemble recommender system can achieve great recommendation performance by effectively combining the decisions generated by individual models. In this paper, we propose a novel ensemble recommender system that combines predictions made by different models into a unified hypergraph ranking framework. This is the first time that hypergraph ranking has been employed to model an ensemble of recommender systems. Hypergraphs are generalizations of graphs where multiple vertices can be connected via hyperedges, efficiently modeling high-order relations. We differentiate real and predicted connections between users and items by assigning different hyperedge weights to individual recommender systems. We perform experiments using four datasets from the fields of movie, music and news media recommendation. The obtained results show that the ensemble hypergraph ranking method generates more accurate recommendations compared to the individual models and a weighted hybrid approach. The assignment of different hyperedge weights to the ensemble hypergraph further improves the performance compared to a setting with identical hyperedge weights.
SemSup-XC: Semantic Supervision for Zero and Few-shot Extreme Classification
Aggarwal, Pranjal, Deshpande, Ameet, Narasimhan, Karthik
Extreme classification (XC) involves predicting over large numbers of classes (thousands to millions), with real-world applications like news article classification and e-commerce product tagging. The zero-shot version of this task requires generalization to novel classes without additional supervision. In this paper, we develop SemSup-XC, a model that achieves state-of-the-art zero-shot and few-shot performance on three XC datasets derived from legal, e-commerce, and Wikipedia data. To develop SemSup-XC, we use automatically collected semantic class descriptions to represent classes and facilitate generalization through a novel hybrid matching module that matches input instances to class descriptions using a combination of semantic and lexical similarity. Trained with contrastive learning, SemSup-XC significantly outperforms baselines and establishes state-of-the-art performance on all three datasets considered, gaining up to 12 precision points on zero-shot and more than 10 precision points on one-shot tests, with similar gains for recall@10. Our ablation studies highlight the relative importance of our hybrid matching module and automatically collected class descriptions.
Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions
Trivedi, Harsh, Balasubramanian, Niranjan, Khot, Tushar, Sabharwal, Ashish
Prompting-based large language models (LLMs) are surprisingly powerful at generating natural language reasoning steps or Chains-of-Thoughts (CoT) for multi-step question answering (QA). They struggle, however, when the necessary knowledge is either unavailable to the LLM or not up-to-date within its parameters. While using the question to retrieve relevant text from an external knowledge source helps LLMs, we observe that this one-step retrieve-and-read approach is insufficient for multi-step QA. Here, \textit{what to retrieve} depends on \textit{what has already been derived}, which in turn may depend on \textit{what was previously retrieved}. To address this, we propose IRCoT, a new approach for multi-step QA that interleaves retrieval with steps (sentences) in a CoT, guiding the retrieval with CoT and in turn using retrieved results to improve CoT. Using IRCoT with GPT3 substantially improves retrieval (up to 21 points) as well as downstream QA (up to 15 points) on four datasets: HotpotQA, 2WikiMultihopQA, MuSiQue, and IIRC. We observe similar substantial gains in out-of-distribution (OOD) settings as well as with much smaller models such as Flan-T5-large without additional training. IRCoT reduces model hallucination, resulting in factually more accurate CoT reasoning. Code, data, and prompts are available at \url{https://github.com/stonybrooknlp/ircot}
What to Do About Fake Drake Songs
On April 3, 2001, Alanis Morissette and Don Henley appeared before Congress in a bid to save the music industry. Henley, the drummer and a lead vocalist for the Eagles, was dressed in a pin-striped suit. Morissette, the Grammy Award-winning singer of "You Oughta Know," wore a red top and a purple ring. Also present was Hilary Rosen, the president and C.E.O. of the Recording Industry Association of America (R.I.A.A.); Shawn Fanning, the co-founder of Napster; Ken Berry, the president and C.E.O. of EMI Recorded Music; and Dianne Feinstein, the then sixty-seven-year-old senator from California. The Senate Judiciary Committee had called the hearing because online file sharing was understood to be threatening the viability of the entire music industry, and of the future of art in America. As the sole musicians to testify, Morissette and Henley might have chosen to echo the chorus of their record-industry colleagues, bemoaning piracy and praising the R.I.A.A.'s moves to stop it.
Anti-woke movie guide site gives conservatives alternative to 'Rotten Tomatoes'
'The Five' co-hosts discuss Rep. Cori Bush, D-Mo., arguing at a congressional hearing that being anti-woke is being anti-Black. Hollywood and movie critics have become "predictable" in pursuing a divisive political agenda in films, a conservative film reviewer says, leaving audiences alienated and frustrated. James Carrick, founder of "Worth it or Woke?", told Fox News Digital he decided to create a movie review site that is transparent about its conservative Christian perspective, as an alternative to long-established film rating sites like "Rotten Tomatoes." The site launched in February and has reviews from current films in theaters, streaming series and favorite films that came out years ago which Carrick believes are "worth it" for audiences to see. A film enthusiast with a background in theater and philosophy, Carrick said he hopes his site will help others "vote with their dollars" to avoiding films with woke messages if they choose.
Ultra-fast boxing robots could be used for real-life fighting game
Remotely operated one-armed boxing robots, inspired by the 2011 movie Real Steel, can respond to a person's movements in as little as one hundredth of a second. "Taking inspiration from the film and incentivised by the entertainment value, we decided to see if we can realise the concept of real life fighting robots," says Yining Geng at the Qibo Robot Company in Weihai, China, which created the robots, dubbed QIBBOTs. Turning the movie's premise into reality required robots that can operate faster than anything previously seen. Remotely operated robots are already used for applications including surgery and seabed exploration, but all have a time-lag or latency, typically 100 milliseconds. This doesn't interfere with the careful, deliberate movements of surgery or remote sample collection, but it does impair the experience of a fast-moving fighting game.
The Morning After: Amazon Prime Day kicks off July 11th
Amazon has announced the dates for its next annual shopping event. Prime Day 2023 will be on July 11th and 12th this year, beginning at 12AM PT/ 3AM ET on Tuesday, July 11th, and concluding at the end of Wednesday, July 12th. Prime Day isn't necessarily a perk of Amazon's subscription service, like access to Prime Video content, but most deals on Amazon during the two-day event are exclusively available to Prime members. The cost of Prime has increased quite a bit since its launch in 2005, and even in the past few years. An annual membership will set you back $139 right now, $20 more than its previous price.
Every 'Black Mirror' Episode, Ranked From Worst to Best
After a four-year hiatus, Black Mirror is back. Season six is now on Netflix, along with the whole back catalog--including one Christmas special and an interactive movie. The show, created by Charlie Brooker and producer Annabel Jones, is a modern take on classic anthology series like The Twilight Zone. Through Brooker's dark, playful, and sometimes uplifting lens, the show examines the unintended ways technology impacts our lives. Because it's an anthology series--in which each installment has new subject matter and a slightly different tone--each episode has its fans.
Edit video like a pro: Absolutely free tools and resources
In some categories of computer software, there are scant free options and spending a few bucks on paid software makes a big difference. In other categories, paying for software is not necessary. Video editing is one such example. In fact, one program widely used in Hollywood is free, and there are several other decent programs for Windows users--including the utterly fantastic DaVinci Resolve. Windows even has a built-in program now that Microsoft is the owner of the simple video editor Clipchamp.
Beyond ChatGPT: AI conspiracy theories are here. Don't believe everything you read.
For as long as there have been scientific breakthroughs and technological innovations, people have been labeling them as magic, witchcraft or the product of nefarious conspiracies directed by powerful, unseen actors. Medieval metalworkers, who transformed stone into jewelry and swords, were seen as agents of either the ruling class or the supernatural, threatening the social fabric. Many still believe that the moon landing was faked in a TV studio. More recently, conspiracy theories that falsely claimed 5G cell technology spread COVID-19 led to attacks on cell towers in the United Kingdom. Artificial intelligence is a technology ready-made for conspiratorial thinking.