Retrieval-Enhanced Machine Learning: Synthesis and Opportunities

In the field of language modeling, models augmented with retrieval components have emerged as a promising solution to address several challenges faced in the natural language processing (NLP) field, including knowledge grounding, interpretability, and scalability. Despite the primary focus on NLP, we posit that the paradigm of retrieval-enhancement can be extended to a broader spectrum of machine learning (ML) such as computer vision, time series prediction, and computational biology. Therefore, this work introduces a formal framework of this paradigm, Retrieval-Enhanced Machine Learning (REML), by synthesizing the literature in various domains in ML with consistent notations which is missing from the current literature. Also, we found that while a number of studies employ retrieval components to augment their models, there is a lack of integration with foundational Information Retrieval (IR) research. We bridge this gap between the seminal IR research and contemporary REML studies by investigating each component that comprises the REML framework. Ultimately, the goal of this work is to equip researchers across various disciplines with a comprehensive, formally structured framework of retrieval-enhanced models, thereby fostering interdisciplinary future research.

Extrinsic Evaluation of Cultural Competence in Large Language Models

Productive interactions between diverse users and language technologies require outputs from the latter to be culturally relevant and sensitive. Prior works have evaluated models' knowledge of cultural norms, values, and artifacts, without considering how this knowledge manifests in downstream applications. In this work, we focus on extrinsic evaluation of cultural competence in two text generation tasks, open-ended question answering and story generation. We quantitatively and qualitatively evaluate model outputs when an explicit cue of culture, specifically nationality, is perturbed in the prompts. Although we find that model outputs do vary when varying nationalities and feature culturally relevant words, we also find weak correlations between text similarity of outputs for different countries and the cultural values of these countries. Finally, we discuss important considerations in designing comprehensive evaluation of cultural competence in user-facing tasks.

Density-based User Representation through Gaussian Process Regression for Multi-interest Personalized Retrieval

Accurate modeling of the diverse and dynamic interests of users remains a significant challenge in the design of personalized recommender systems. Existing user modeling methods, like single-point and multi-point representations, have limitations w.r.t. accuracy, diversity, computational cost, and adaptability. To overcome these deficiencies, we introduce density-based user representations (DURs), a novel model that leverages Gaussian process regression for effective multi-interest recommendation and retrieval. Our approach, GPR4DUR, exploits DURs to capture user interest variability without manual tuning, incorporates uncertainty-awareness, and scales well to large numbers of users. Experiments using real-world offline datasets confirm the adaptability and efficiency of GPR4DUR, while online experiments with simulated users demonstrate its ability to address the exploration-exploitation trade-off by effectively utilizing model uncertainty.

Distributionally-Informed Recommender System Evaluation

Current practice for evaluating recommender systems typically focuses on point estimates of user-oriented effectiveness metrics or business metrics, sometimes combined with additional metrics for considerations such as diversity and novelty. In this paper, we argue for the need for researchers and practitioners to attend more closely to various distributions that arise from a recommender system (or other information access system) and the sources of uncertainty that lead to these distributions. One immediate implication of our argument is that both researchers and practitioners must report and examine more thoroughly the distribution of utility between and within different stakeholder groups. However, distributions of various forms arise in many more aspects of the recommender systems experimental process, and distributional thinking has substantial ramifications for how we design, evaluate, and present recommender systems evaluation and research results. Leveraging and emphasizing distributions in the evaluation of recommender systems is a necessary step to ensure that the systems provide appropriate and equitably-distributed benefit to the people they affect.

Fairness Through Domain Awareness: Mitigating Popularity Bias For Music Discovery

As online music platforms grow, music recommender systems play a vital role in helping users navigate and discover content within their vast musical databases. At odds with this larger goal, is the presence of popularity bias, which causes algorithmic systems to favor mainstream content over, potentially more relevant, but niche items. In this work we explore the intrinsic relationship between music discovery and popularity bias. To mitigate this issue we propose a domain-aware, individual fairness-based approach which addresses popularity bias in graph neural network (GNNs) based recommender systems. Our approach uses individual fairness to reflect a ground truth listening experience, i.e., if two songs sound similar, this similarity should be reflected in their representations. In doing so, we facilitate meaningful music discovery that is robust to popularity bias and grounded in the music domain. We apply our BOOST methodology to two discovery based tasks, performing recommendations at both the playlist level and user level. Then, we ground our evaluation in the cold start setting, showing that our approach outperforms existing fairness benchmarks in both performance and recommendation of lesser-known content. Finally, our analysis explains why our proposed methodology is a novel and promising approach to mitigating popularity bias and improving the discovery of new and niche content in music recommender systems.

Group Membership Bias

When learning to rank from user interactions, search and recommendation systems must address biases in user behavior to provide a high-quality ranking. One type of bias that has recently been studied in the ranking literature is when sensitive attributes, such as gender, have an impact on a user's judgment about an item's utility. For example, in a search for an expertise area, some users may be biased towards clicking on male candidates over female candidates. We call this type of bias group membership bias or group bias for short. Increasingly, we seek rankings that not only have high utility but are also fair to individuals and sensitive groups. Merit-based fairness measures rely on the estimated merit or utility of the items. With group bias, the utility of the sensitive groups is under-estimated, hence, without correcting for this bias, a supposedly fair ranking is not truly fair. In this paper, first, we analyze the impact of group bias on ranking quality as well as two well-known merit-based fairness metrics and show that group bias can hurt both ranking and fairness. Then, we provide a correction method for group bias that is based on the assumption that the utility score of items in different groups comes from the same distribution. This assumption has two potential issues of sparsity and equality-instead-of-equity, which we use an amortized approach to solve. We show that our correction method can consistently compensate for the negative impact of group bias on ranking quality and fairness metrics.

Scaling Laws Do Not Scale

Recent work has proposed a power law relationship, referred to as ``scaling laws,'' between the performance of artificial intelligence (AI) models and aspects of those models' design (e.g., dataset size). In other words, as the size of a dataset (or model parameters, etc) increases, the performance of a given model trained on that dataset will correspondingly increase. However, while compelling in the aggregate, this scaling law relationship overlooks the ways that metrics used to measure performance may be precarious and contested, or may not correspond with how different groups of people may perceive the quality of models' output. In this paper, we argue that as the size of datasets used to train large AI models grows, the number of distinct communities (including demographic groups) whose data is included in a given dataset is likely to grow, each of whom may have different values. As a result, there is an increased risk that communities represented in a dataset may have values or preferences not captured by (or in the worst case, at odds with) the metrics used to evaluate model performance for scaling laws. We end the paper with implications for AI scaling laws -- that models may not, in fact, continue to improve as the datasets get larger -- at least not for all people or communities impacted by those models.

Best-Case Retrieval Evaluation: Improving the Sensitivity of Reciprocal Rank with Lexicographic Precision

Across a variety of ranking tasks, researchers use reciprocal rank to measure the effectiveness for users interested in exactly one relevant item. Despite its widespread use, evidence suggests that reciprocal rank is brittle when discriminating between systems. This brittleness, in turn, is compounded in modern evaluation settings where current, high-precision systems may be difficult to distinguish. We address the lack of sensitivity of reciprocal rank by introducing and connecting it to the concept of best-case retrieval, an evaluation method focusing on assessing the quality of a ranking for the most satisfied possible user across possible recall requirements. This perspective allows us to generalize reciprocal rank and define a new preference-based evaluation we call lexicographic precision or lexiprecision. By mathematical construction, we ensure that lexiprecision preserves differences detected by reciprocal rank, while empirically improving sensitivity and robustness across a broad set of retrieval and recommendation tasks.

Commonality in Recommender Systems: Evaluating Recommender Systems to Enhance Cultural Citizenship

Recommender systems have become the dominant means of curating cultural content, significantly influencing individual cultural experience. Since recommender systems tend to optimize for personalized user experience, they can overlook impacts on cultural experience in the aggregate. After demonstrating that existing metrics do not center culture, we introduce a new metric, commonality, that measures the degree to which recommendations familiarize a given user population with specified categories of cultural content. We developed commonality through an interdisciplinary dialogue between researchers in computer science and the social sciences and humanities. With reference to principles underpinning public service media systems in democratic societies, we identify universality of address and content diversity in the service of strengthening cultural citizenship as particularly relevant goals for recommender systems delivering cultural content. We develop commonality as a measure of recommender system alignment with the promotion of content toward a shared cultural experience across a population of users. We empirically compare the performance of recommendation algorithms using commonality with existing metrics, demonstrating that commonality captures a novel property of system behavior complementary to existing metrics. Alongside existing fairness and diversity metrics, commonality contributes to a growing body of scholarship developing `public good' rationales for machine learning systems.

Recall as a Measure of Ranking Robustness

Researchers use recall to evaluate rankings across a variety of retrieval, recommendation, and machine learning tasks. While there is a colloquial interpretation of recall in set-based evaluation, the research community is far from a principled understanding of recall metrics for rankings. The lack of principled understanding of or motivation for recall has resulted in criticism amongst the retrieval community that recall is useful as a measure at all. In this light, we reflect on the measurement of recall in rankings from a formal perspective. Our analysis is composed of three tenets: recall, robustness, and lexicographic evaluation. First, we formally define `recall-orientation' as sensitivity to movement of the bottom-ranked relevant item. Second, we analyze our concept of recall orientation from the perspective of robustness with respect to possible searchers and content providers. Finally, we extend this conceptual and theoretical treatment of recall by developing a practical preference-based evaluation method based on lexicographic comparison. Through extensive empirical analysis across 17 TREC tracks, we establish that our new evaluation method, lexirecall, is correlated with existing recall metrics and exhibits substantially higher discriminative power and stability in the presence of missing labels. Our conceptual, theoretical, and empirical analysis substantially deepens our understanding of recall and motivates its adoption through connections to robustness and fairness.