Deep learning techniques have become the method of choice for researchers working on algorithmic aspects recommender systems. With strongly increased interest in machine general, it has, as a result, difficult to keep track what represents state-of-the-art at moment, e.g., top-n recommendation tasks. At same time, several recent publications point out problems today's research practice applied learning, terms reproducibility results or baselines when proposing new models.

The design of algorithms that generate personalized ranked item lists is a central topic research in the field recommender systems. In past few years, particular, approaches based on deep learning (neural) techniques have become dominant literature. For all them, substantial progress over state-of-the-art claimed. However, indications exist certain problems today's practice, e.g., with respect to choice and optimization baselines used for comparison, raising questions about published claims....

As of today, most movie recommendation services base their recommendations on collaborative filtering (CF) and/or content-based (CBF) models that use metadata (e.g., genre or cast). In video-on-demand and streaming services, however, new movies TV series are continuously added. CF unable to make predictions in such a scenario, since the newly added videos lack interactions—a problem technically known as item cold start (CS). Currently, common approach this is switch purely CBF method,...

Job Shop Scheduling is a combinatorial optimization problem of particular importance for production environments where the goal to complete task in shortest possible time given limitations resources available. Due its computational complexity it quickly becomes intractable problems interesting size. The emerging technology Quantum Annealing provides an alternative architecture that promises improved scalability and solution quality. However, several as well open research questions exist this...

Feature selection is a common step in many ranking, classification, or prediction tasks and serves purposes. By removing redundant noisy features, the accuracy of ranking classification can be improved computational cost subsequent learning steps reduced. However, feature itself computationally expensive process. While for decades confined to theoretical algorithmic papers, quantum computing now becoming viable tool tackle realistic problems, particular special-purpose solvers based on...

Novel data sources bring new opportunities to improve the quality of recommender systems and serve as a catalyst for creation paradigms on personalized recommendations. Impressions are novel source containing items shown users their screens. Past research focused providing recommendations using interactions, occasionally impressions when such was available. Interest in has increased due potential provide more accurate Despite this interest, is still dispersed. Many works have distinct...

Lately, we have observed a growing interest in intent-aware recommender systems (IARS). The promise of such is that they are capable generating better recommendations by predicting and considering the underlying motivations short-term goals consumers. From technical perspective, various sophisticated neural models were recently proposed this emerging promising area. In broader context complex recommendation models, number research works unfortunately indicates (i) reproducing often difficult...

In this paper, we discuss the Walsh Series Loader (WSL) algorithm, proposed by. particular, observe that paper does not describe how to implement term of order zero operator WSL is based on. While affect theoretical validity WSL, it poses obstacles for practitioners aiming use because, as show in our experiments, an incorrect implementation leads states with very poor fidelity. full quantum circuit required by including zero, releasing source code online, and algorithm works correctly....

Feature selection is of great importance in Machine Learning, where it can be used to reduce the dimensionality classification, ranking and prediction problems. The removal redundant noisy features improve both accuracy scalability trained models. However, feature a computationally expensive task with solution space that grows combinatorically. In this work, we consider particular quadratic problem tackled Quantum Approximate Optimization Algorithm (QAOA), already employed combinatorial...

In this article, we introduce the ContentWise Impressions dataset, a collection of implicit interactions and impressions movies TV series from an Over-The-Top media service, which delivers its contents over Internet. The dataset is distinguished other already available multimedia recommendation datasets by availability impressions, i.e., recommendations shown to user, size, being open-source. We describe data process, preprocessing applied, characteristics, statistics when compared commonly...

In this paper we provide an overview of the approach used as team Creamy Fireflies for ACM RecSys Challenge 2018. The competition, organized by Spotify, focuses on problem playlist continuation, that is suggesting which tracks user may add to existing playlist. challenge addresses issue in many use cases, from cold start playlists already composed up a hundred tracks. Our proposes solution based few well known models both content and collaborative, whose predictions are aggregated via...

After decades of being mainly confined to theoretical research, Quantum Computing is now becoming a useful tool for solving realistic problems. This work aims experimentally explore the feasibility using currently available quantum computers, based on Annealing paradigm, build recommender system exploiting community detection. Community detection, by partitioning users and items into densely connected clusters, can boost accuracy non-personalized recommendation assuming that within each...

In Recommender Systems, impressions are a relatively new type of information that records all products previously shown to the users. They also complex source information, combining effects recommender system generated them, search results, or business rules may select specific for recommendations. The fact user interacted with item given list recommended ones benefit from richer interaction signal, in which some items did not interact be considered negative interactions. This work presents...

It has been long known that quantum computing the potential to revolutionize way we find solutions of problems are difficult solve on classical computers. was only recently small but functional computers have become available cloud, allowing test their potential. In this paper propose leverage capabilities address an important task for recommender systems providers, optimal selection recommendation carousels. many video-on-demand and music streaming services user is provided with a homepage...

In this paper we provide a description of the methods used as team BanaNeverAlone for ACM RecSys Challenge 2020, organized by Twitter. The challenge addresses problem user engagement prediction: goal is to predict probability (Like, Reply, Retweet or with comment), based on series past interactions Twitter platform. Our proposed solution relies several features that extracted from original dataset, well consolidated models, such gradient boosting decision trees and neural networks. ensemble...

Many video-on-demand and music streaming services provide the user with a page consisting of several recommendation lists, i.e., widgets or swipeable carousels , each built specific criteria (e.g., most recent, TV series, etc.). Finding efficient strategies to select which display is an active research topic great industrial interest. In this setting, overall quality recommendations new algorithm cannot be assessed by measuring solely its individual quality. Rather, it should evaluated in...

This paper focuses on recommender systems based item-item collaborative filtering (CF). Although research item-based methods is not new, current literature does provide any reliable insight how to estimate confidence of recommendations. The goal this fill gap, by investigating the conditions under which recommendations will succeed or fail for a specific user. We formalize CF problem as an eigenvalue problem, where estimated ratings are equivalent true (unknown) multiplied user-specific...

The promise of quantum computing to open new unexplored possibilities in several scientific fields has been long discussed, but until recently the lack a functional computer confined this discussion mostly theoretical algorithmic papers. It was only last few years that small computers have become available broader research community. One paradigm particular, annealing, can be used sample optimal solutions for number NP-hard optimization problems represented with classical operations tools,...

In this paper we provide an overview of the approach used as team PoliCloud8 for ACM RecSys Challenge 2019. The competition, organized by Trivago, focuses on problem session-based and context-aware accommodation recommendation in a travel domain. goal is to suggest suitable accommodations fitting needs traveller maximise chance redirect (click-out) booking site, relying explicit implicit user signals within session (clicks, search refinement, filter usage) detect users intent. Our proposes...