Extensive literature spanning psychology, sociology, and musicology has sought to understand the motivations for why people listen to music, including both individually and socially motivated reasons. Music's social functions, while present throughout the world, may be particularly important in collectivist societies, but music recommender systems generally target individualistic functions of music listening. In this study, we explore how a recommender system focused on social motivations for music listening might work by addressing a particular motivation: the desire to listen to music that is trending in one’s community. We frame a recommendation task suited to this desire and propose a corresponding evaluation metric to address the timeliness of recommendations. Using listening data from Spotify, we construct a simple, heuristic-based approach to introduce and explore this recommendation task. Analyzing the effectiveness of this approach, we discuss what we believe is an overlooked trade-off between the precision and timeliness of recommendations, as well as considerations for modeling users' musical communities. Finally, we highlight key cultural differences in the effectiveness of this approach, underscoring the importance of incorporating a diverse cultural perspective in the development and evaluation of recommender systems.

Recommender systems are ubiquitous and influence the information we consume daily by helping us navigate vast catalogs of information like music databases. However, their linear approach of surfacing content in ranked lists limits their ability to help us grow and understand our personal preferences. In this paper, we study how we can better support users in exploring a novel space, specifically focusing on music genres. Informed by interviews with expert music listeners, we developed TastePaths: an interactive web tool that helps users explore an overview of the genre-space via a graph of connected artists. We conducted a comparative user study with 16 participants where each of them used a personalized version of TastePaths (built with a set of artists the user listens to frequently) and a non-personalized one (based on a set of the most popular artists in a genre). We find that participants employed various strategies to explore the space. Overall, they greatly preferred the personalized version as it helped anchor their exploration and provided recommendations that were more compatible with their personal taste. In addition to that, TastePaths helped participants specify and articulate their interest in the genre and gave them a better understanding of the system’s organization of music. Based on our findings, we discuss opportunities and challenges for incorporating more control and expressive feedback in recommendation systems to help users explore spaces beyond their immediate interests and improve these systems’ underlying algorithms.

In order to support fairness-forward thinking by machine learning (ML) practitioners, fairness researchers have created toolkits that aim to transform state-of-the-art research contributions into easily-accessible APIs. Despite these efforts, recent research indicates a disconnect between the needs of practitioners and the tools offered by fairness research. By engaging 20 ML practitioners in a simulated scenario in which they utilize fairness toolkits to make critical decisions, this work aims to utilize practitioner feedback to inform recommendations for the design and creation of fair ML toolkits. Through the use of survey and interview data, our results indicate that though fair ML toolkits are incredibly impactful on users’ decision-making, there is much to be desired in the design and demonstration of fairness results. To support the future development and evaluation of toolkits, this work offers a rubric that can be used to identify critical components of Fair ML toolkits.

Women are dramatically underrepresented in computer science at all levels in academia and account for just 15% of tenure-track faculty. Understanding the causes of this gender imbalance would inform both policies intended to rectify it and employment decisions by departments and individuals. Progress in this direction, however, is complicated by the complexity and decentralized nature of faculty hiring and the non-independence of hires. Using comprehensive data on both hiring outcomes and scholarly productivity for 2659 tenure-track faculty across 205 Ph.D.-granting departments in North America, we investigate the multi-dimensional nature of gender inequality in computer science faculty hiring through a network model of the hiring process. Overall, we find that hiring outcomes are most directly affected by (i) the relative prestige between hiring and placing institutions and (ii) the scholarly productivity of the candidates. After including these, and other features, the addition of gender did not significantly reduce modeling error. However, gender differences do exist, e.g., in scholarly productivity, postdoctoral training rates, and in career movements up the rankings of universities, suggesting that the effects of gender are indirectly incorporated into hiring decisions through gender's covariates. Furthermore, we find evidence that more highly ranked departments recruit female faculty at higher than expected rates, which appears to inhibit similar efforts by lower ranked departments. These findings illustrate the subtle nature of gender inequality in faculty hiring networks and provide new insights to the underrepresentation of women in computer science.

BACKGROUND: Computational analysis of metagenomes requires the taxonomical assignment of the genome contigs assembled from DNA reads of environmental samples. Because of the diverse nature of microbiomes, the length of the assemblies obtained can vary between a few hundred bp to a few hundred Kbp. Current taxonomic classification algorithms provide accurate classification for long contigs or for short fragments from organisms that have close relatives with annotated genomes. These are significant limitations for metagenome analysis because of the complexity of microbiomes and the paucity of existing annotated genomes. RESULTS: We propose a robust taxonomic classification method, RAIphy, that uses a novel sequence similarity metric with iterative refinement of taxonomic models and functions effectively without these limitations. We have tested RAIphy with synthetic metagenomics data ranging between 100 bp to 50 Kbp. Within a sequence read range of 100 bp-1000 bp, the sensitivity of RAIphy ranges between 38%-81% outperforming the currently popular composition-based methods for reads in this range. Comparison with computationally more intensive sequence similarity methods shows that RAIphy performs competitively while being significantly faster. The sensitivity-specificity characteristics for relatively longer contigs were compared with the PhyloPythia and TACOA algorithms. RAIphy performs better than these algorithms at varying clade-levels. For an acid mine drainage (AMD) metagenome, RAIphy was able to taxonomically bin the sequence read set more accurately than the currently available methods, Phymm and MEGAN, and more accurately in two out of three tests than the much more computationally intensive method, PhymmBL. CONCLUSIONS: With the introduction of the relative abundance index metric and an iterative classification method, we propose a taxonomic classification algorithm that performs competitively for a large range of DNA contig lengths assembled from metagenome data. Because of its speed, simplicity, and accuracy RAIphy can be successfully used in the binning process for a broad range of metagenomic data obtained from environmental samples.

BACKGROUND: We propose a sequence clustering algorithm and compare the partition quality and execution time of the proposed algorithm with those of a popular existing algorithm. The proposed clustering algorithm uses a grammar-based distance metric to determine partitioning for a set of biological sequences. The algorithm performs clustering in which new sequences are compared with cluster-representative sequences to determine membership. If comparison fails to identify a suitable cluster, a new cluster is created. RESULTS: The performance of the proposed algorithm is validated via comparison to the popular DNA/RNA sequence clustering approach, CD-HIT-EST, and to the recently developed algorithm, UCLUST, using two different sets of 16S rDNA sequences from 2,255 genera. The proposed algorithm maintains a comparable CPU execution time with that of CD-HIT-EST which is much slower than UCLUST, and has successfully generated clusters with higher statistical accuracy than both CD-HIT-EST and UCLUST. The validation results are especially striking for large datasets. CONCLUSIONS: We introduce a fast and accurate clustering algorithm that relies on a grammar-based sequence distance. Its statistical clustering quality is validated by clustering large datasets containing 16S rDNA sequences.