Recommendation and ranking systems are known to suffer from popularity bias; the tendency of the algorithm to favor a few popular items while under-representing the majority of other items. Prior research has examined various approaches for mitigating popularity bias and enhancing the recommendation of long-tail, less popular, items. The effectiveness of these approaches is often assessed using different metrics to evaluate the extent to which over-concentration on popular items is reduced. However, not much attention has been given to the user-centered evaluation of this bias; how different users with different levels of interest towards popular items are affected by such algorithms. In this paper, we show the limitations of the existing metrics to evaluate popularity bias mitigation when we want to assess these algorithms from the users' perspective and we propose a new metric that can address these limitations. In addition, we present an effective approach that mitigates popularity bias from the user-centered point of view. Finally, we investigate several state-of-the-art approaches proposed in recent years to mitigate popularity bias and evaluate their performances using the existing metrics and also from the users' perspective. Our experimental results using two publicly-available datasets show that existing popularity bias mitigation techniques ignore the users' tolerance towards popular items. Our proposed user-centered method can tackle popularity bias effectively for different users while also improving the existing metrics.

Collaborative Filtering (CF) based recommendation methods have been widely studied, which can be generally categorized into two types, i.e., representation learning-based CF methods and matching function learning-based CF methods. Representation learning tries to learn a common low dimensional space for the representations of users and items. In this case, a user and item match better if they have higher similarity in that common space. Matching function learning tries to directly learn the complex matching function that maps user-item pairs to matching scores. Although both methods are well developed, they suffer from two fundamental flaws, i.e., the representation learning resorts to applying a dot product which has limited expressiveness on the latent features of users and items, while the matching function learning has weakness in capturing low-rank relations. To overcome such flaws, we propose a novel recommendation model named Balanced Collaborative Filtering Network (BCFNet), which has the strengths of the two types of methods. In addition, an attention mechanism is designed to better capture the hidden information within implicit feedback and strengthen the learning ability of the neural network. Furthermore, a balance module is designed to alleviate the over-fitting issue in DNNs. Extensive experiments on eight real-world datasets demonstrate the effectiveness of the proposed model.

Many practical recommender systems provide item recommendation for different users only via mining user-item interactions but totally ignoring the rich attribute information of items that users interact with. In this paper, we propose an attribute-augmented graph neural network model named Murzim. Murzim takes as input the graphs constructed from the user-item interaction sequences and corresponding item attribute sequences. By combining the GNNs with node aggregation and an attention network, Murzim can capture user preference patterns, generate embeddings for user-item interaction sequences, and then generate recommendations through next-item prediction. We conduct extensive experiments on multiple datasets. Experimental results show that Murzim outperforms several state-of-the-art methods in terms of recall and MRR, which illustrates that Murzim can make use of item attribute information to produce better recommendations. At present, Murzim has been deployed in MX Player, one of India's largest streaming platforms, and is recommending videos for tens of thousands of users.

The wireless home-audio specialist Sonos has unveiled the Roam, its smaller, cheaper portable speaker with Bluetooth and wifi that works as well at home as it does outdoors. The rugged triangular Roam weighs 430g and is about the size of a water bottle. The aim is that it will avoid ending up stuck in a drawer collecting dust like most portable speakers by sounding good enough and working well enough to warrant being used at home too, connecting to Sonos' smart multi-room wifi system. That means it will work like any other non-battery powered Sonos speaker able to stream music directly via wifi from more than 100 different music services, including Spotify and Apple Music, and be grouped with other speakers to play music all over the home. It also supports Apple's AirPlay 2 and smart speaker functionality with either Google Assistant and Amazon's Alexa. The Roam has separate woofer and tweeter speakers with Sonos's Trueplay technology, which automatically tunes the sound accounting for acoustics, obstacles and position to sound its best at all times.

As impressive as Amazon's Alexa technology is, the best way to utilize it isn't always clear. Sure, you can get it to tell you the weather, play music and set alarms, but that's just scratching the surface of what this AI is truly capable of. And if you never get around to purchasing and syncing compatible peripherals, you won't get to explore much further than that. If you haven't quite been sold on voice-activated tech yet, or simply want to up the ante a spell, it's absolutely worth checking out the GE C by GE Sol, an Alexa-enabled smart light that's available right now for just $90 (normally $199). It has Amazon Alexa built right into it, so it serves a direct function in addition to having such a charming voice.

Cortana is continuing its slow fade with word that the Harman Kardon Invoke, the final smart speaker powered by Microsoft's once-promising voice assistant, will disable it with an impending update. Slated for release on Wednesday, the previously announced update will essentially turn the Invoke into a "dumb" Bluetooth speaker, Paul Thurrott reports. An FAQ on Karman Kardon's support site says the update will install itself silently in the background over Wi-Fi. Once the update is applied, the speaker will no longer respond to "Hey Cortana," nor will it connect to Wi-Fi anymore. Microsoft says it will give anyone with an "active" Invoke a $50 gift card in compensation for the lost functionality.

Harman Kardon's Invoke was the first speaker to use Microsoft's Cortana speaker, and as it later turned out, also the last. In 2020, Microsoft said it would end consumer support for Cortana on iOS, Android and Invoke by early 2021 and use the AI assistant exclusively for business apps like Microsoft 365. That day has now come, as Harman Kardon has announced that it will soon release a software update that disables Cortana and turns the Invoke into a regular Bluetooth speaker, Thurott has reported. The update will arrive tomorrow, so in a FAQ, Harman Kardon advised owners to leave the speaker "connected to the internet and kept online overnight between 2 AM and 5 AM local time." The update will end after June 30th, 2021, but the Samsung-owned company noted that "Cortana service... will end in the coming months regardless of whether you receive the update."

Amazon's Echo and other smart speakers like the Google Home could be used to monitor the rhythm of a person's heart. Academics created an AI-powered device which monitors regular, and irregular, heartbeats using the same tools found in smart speakers. The prototype, which was built in a lab but could be incorporated into speakers in the future, was found to be almost as good as medical devices in hospitals. The search for heartbeats begins when a person sits within one to two feet of the smart speaker. Then the system plays an inaudible continuous sound, which bounces off the person and then returns to the speaker.

This resource is continuously updated. If you know any other suitable and open datasets, please let us know by emailing us at pub@towardsai.net or by dropping a comment below. Google Dataset Search: Similar to how Google Scholar works, Dataset Search lets you find datasets wherever they are hosted, whether it's a publisher's site, a digital library, or an author's web page. It's a phenomenal dataset finder, and it contains over 25 million datasets. Kaggle: Kaggle provides a vast container of datasets, sufficient for the enthusiast to the expert.

It responded almost instantly -- a capability that to this day, 5 years and change on, I find miraculous. Before the Amazon Echo, my prior magical Voice First moment of bliss was the launch of Siri on October 4th, 2011, with the release of the 4S. Up to that point, the only way to use speech on your smartphone was through an app. The Siri app was such an app, and I had found it impressive in what it could do. But now, with Apple pulling the app (actually a small apportion of the app) into the iPhone -- a mainstream product that was going places -- well now, speech had finally arrived! Yes, speech had finally arrived, and it had arrived and was being welcomed in the house of the iPhone, where it was given a spacious room to lodge in, alongside the other rooms – the camera, the phone, the texting, etc.