Tip-of-the-Tongue (ToT) retrieval benchmarks have largely focused on English, limiting their applicability to multilingual information access. In this work, we construct multilingual ToT test collections for Chinese, Japanese, Korean, and English, using an LLM-based query simulation framework. We systematically study how prompt language and source document language affect the fidelity of simulated ToT queries, validating synthetic queries through system rank correlation against real user queries. Our results show that effective ToT simulation requires language-aware design choices: non-English language sources are generally important, while English Wikipedia can be beneficial when non-English sources provide insufficient information for query generation. Based on these findings, we release four ToT test collections with 5,000 queries per language across multiple domains. This work provides the first large-scale multilingual ToT benchmark and offers practical guidance for constructing realistic ToT datasets beyond English.

Retrieval-Augmented Generation (RAG) systems typically rely on a single fixed retriever, despite growing evidence that no single retriever performs optimally across all query types. In this paper, we explore a query routing approach that dynamically selects from a pool of retrievers based on the query, using both train-free heuristics and learned routing models. We frame routing as a learning-to-rank problem and introduce LTRR, a framework that Learns To Rank Retrievers according to their expected contribution to downstream RAG performance. Through experiments on diverse question-answering benchmarks with controlled variations in query types, we demonstrate that routing-based RAG consistently surpasses the strongest single-retriever baselines. The gains are particularly substantial when training with the Answer Correctness (AC) objective and when using pairwise ranking methods, with XGBoost yielding the best results. Additionally, our approach exhibits stronger generalization to out-of-distribution queries. Overall, our results underscore the critical role of both training strategy and optimization metric choice in effective query routing for RAG systems.

Modern information access ecosystems consist of mixtures of systems, such as retrieval systems and large language models, and increasingly rely on marketplaces to mediate access to models, tools, and data, making competition between systems inherent to deployment. In such settings, outcomes are shaped not only by benchmark quality but also by competitive pressure, including user switching, routing decisions, and operational constraints. Yet evaluation is still largely conducted on static benchmarks with accuracy-focused measures that assume systems operate in isolation. This mismatch makes it difficult to predict post-deployment success and obscures competitive effects such as early-adoption advantages and market dominance. We introduce Marketplace Evaluation, a simulation-based paradigm that evaluates information access systems as participants in a competitive marketplace. By simulating repeated interactions and evolving user and agent preferences, the framework enables longitudinal evaluation and marketplace-level metrics, such as retention and market share, that complement and can extend beyond traditional accuracy-based metrics. We formalize the framework and outline a research agenda, motivated by business and economics, around marketplace simulation, metrics, optimization, and adoption in evaluation campaigns like TREC.

Users tend to remember failures of a search session more than its many successes. This observation has led to work on search robustness, where systems are penalized if they perform very poorly on some queries. However, this principle of robustness has been overlooked within a single query. An ambiguous or underspecified query (e.g., ''jaguar'') can have several user intents, where popular intents often dominate the ranking, leaving users with minority intents unsatisfied. Although the diversification literature has long recognized this issue, existing metrics only model the average relevance across intents and provide no robustness guarantees. More surprisingly, we show theoretically and empirically that many well-known diversification algorithms are no more robust than a naive, non-diversified algorithm. To address this critical gap, we propose to frame diversification as a risk-minimization problem. We introduce VRisk, which measures the expected risk faced by the least-served fraction of intents in a query. Optimizing VRisk produces a robust ranking, reducing the likelihood of poor user experiences. We then propose VRisker, a fast greedy re-ranker with provable approximation guarantees. Finally, experiments on NTCIR INTENT-2, TREC Web 2012, and MovieLens show the vulnerability of existing methods. VRisker reduces worst-case intent failures by up to 33% with a minimal 2% drop in average performance.

Although originally developed to evaluate sets of items, recall is often used to evaluate rankings of items, including those produced by recommender, retrieval, and other machine learning systems. The application of recall without a formal evaluative motivation has led to criticism of recall as a vague or inappropriate measure. In light of this debate, 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 the sensitivity of a metric to a user interested in finding every relevant item. Second, we analyze recall orientation from the perspective of robustness with respect to possible content consumers and providers, connecting recall to recent conversations about fair ranking. 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 multiple recommendation and retrieval tasks, we establish that our new evaluation method, lexirecall, has convergent validity (i.e., it is correlated with existing recall metrics) and exhibits substantially higher sensitivity in terms of 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.

Retrieval-augmented Generation (RAG) is powerful, but its effectiveness hinges on which retrievers we use and how.Different retrievers offer distinct, often complementary signals: BM25 captures lexical matches; dense retrievers, semantic similarity.Yet in practice, we typically fix a single retriever based on heuristics, which fails to generalize across diverse information needs.Can we dynamically select and integrate multiple retrievers for each individual query, without the need for manual selection?In our work, we validate this intuition with quantitative analysis and introduce a mixture of retrievers: a zero-shot, weighted combination of heterogeneous retrievers.Extensive experiments show that such mixtures are effective and efficient: Despite totaling just 0.8B parameters, this mixture outperforms every individual retriever and even larger 7B models-by +10.8% and +3.9% on average, respectively.Further analysis also shows that this mixture framework can help incorporate specialized non-oracle human information sources as retrievers to achieve good collaboration, with a 58.9% relative performance improvement over simulated humans alone.

In AI research and practice, rigor remains largely understood in terms of methodological rigor---such as whether mathematical, statistical, or computational methods are correctly applied. We argue that this narrow conception of rigor has contributed to the concerns raised by the responsible AI community, including overblown claims about the capabilities of AI systems. Our position is that a broader conception of what rigorous AI research and practice should entail is needed. We believe such a conception---in addition to a more expansive understanding of 1) methodological rigor---should include aspects related to 2) what background knowledge informs what to work on (epistemic rigor); 3) how disciplinary, community, or personal norms, standards, or beliefs influence the work (normative rigor); 4) how clearly articulated the theoretical constructs under use are (conceptual rigor); 5) what is reported and how (reporting rigor); and 6) how well-supported the inferences from existing evidence are (interpretative rigor). In doing so, we also provide useful language and a framework for much needed dialogue about the AI community's work by researchers, policymakers, journalists, and other stakeholders.

Retrieval-Enhanced Machine Learning (REML) refers to the use of information retrieval (IR) methods to support reasoning and inference in machine learning tasks. Although relatively recent, these approaches can substantially improve model performance. This includes improved generalization, knowledge grounding, scalability, freshness, attribution, interpretability, and on-device learning. To date, despite being influenced by work in the information retrieval community, REML research has predominantly been presented in natural language processing (NLP) conferences. Our tutorial addresses this disconnect by introducing core REML concepts and synthesizing the literature from various domains in machine learning (ML), including, but not limited to, NLP. What is unique to our approach is the use of consistent notations to provide researchers with a unified and expandable framework. The tutorial will be presented in lecture format based on an existing manuscript, with supporting materials and a comprehensive reading list available at a website. Building on the momentum of our successful workshop at SIGIR 2023 and our tutorial at SIGIR-AP 2024, this year's tutorial features updated content with an emphasis on retrieval technologies used across the broader ML community. We also highlight their role in emerging, future-facing applications such as language agents and evolving scenarios where the extensive body of knowledge from IR can provide critical insights and capabilities.

Tip-of-the-tongue (TOT) search occurs when a user struggles to recall a specific identifier, such as a document title. While common, existing search systems often fail to effectively support TOT scenarios. Research on TOT retrieval is further constrained by the challenge of collecting queries, as current approaches rely heavily on community question-answering (CQA) websites, leading to labor-intensive evaluation and domain bias. To overcome these limitations, we introduce two methods for eliciting TOT queries-leveraging large language models (LLMs) and human participants-to facilitate simulated evaluations of TOT retrieval systems. Our LLM-based TOT user simulator generates synthetic TOT queries at scale, achieving high correlations with how CQA-based TOT queries rank TOT retrieval systems when tested in the Movie domain. Additionally, these synthetic queries exhibit high linguistic similarity to CQA-derived queries. For human-elicited queries, we developed an interface that uses visual stimuli to place participants in a TOT state, enabling the collection of natural queries. In the Movie domain, system rank correlation and linguistic similarity analyses confirm that human-elicited queries are both effective and closely resemble CQA-based queries. These approaches reduce reliance on CQA-based data collection while expanding coverage to underrepresented domains, such as Landmark and Person. LLM-elicited queries for the Movie, Landmark, and Person domains have been released as test queries in the TREC 2024 TOT track, with human-elicited queries scheduled for inclusion in the TREC 2025 TOT track. Additionally, we provide source code for synthetic query generation and the human query collection interface, along with curated visual stimuli used for eliciting TOT queries.

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, and adaptability. To overcome these deficiencies, we introduce density-based user representations (DURs), a novel method that leverages Gaussian process regression (GPR) 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.

When learning to rank from user interactions, search and recommender 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.

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 article, 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.

Recommender systems have become the dominant means of curating cultural content, significantly influencing the nature of individual cultural experience. While the majority of academic and industrial research on recommender systems optimizes for personalized user experience, this paradigm does not capture the ways that recommender systems impact cultural experience in the aggregate, across populations of users. Although existing novelty, diversity, and fairness studies probe how recommender systems relate to the broader social role of cultural content, they do not adequately center culture as a core concept and challenge. In this work, we introduce commonality as a new measure of recommender systems that reflects the degree to which recommendations familiarize a given user population with specified categories of cultural content. Our proposed commonality metric responds to a set of arguments developed through an interdisciplinary dialogue between researchers in computer science and the social sciences and humanities. With reference to principles underpinning public service media (PSM) 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 a shared cultural experience of, and exposure to, diverse cultural content across a population of users. Moreover, we advocate for the involvement of human editors accountable to a larger value community as a fundamental part of defining categories in the service of cultural citizenship. We empirically compare the performance of recommendation algorithms using commonality with existing utility, diversity, novelty, and fairness metrics using three different domains. Our results demonstrate that commonality captures a property of system behavior complementary to existing metrics and suggests the need for alternative, non-personalized interventions in recommender systems oriented to strengthening cultural citizenship across populations of users. Moreover, commonality demonstrates both consistent results under different editorial policies and robustness to missing labels and users. Alongside existing fairness and diversity metrics, commonality contributes to a growing body of scholarship developing “public good” rationales for digital media and machine learning systems.

Most machine learning models are designed to be self-contained and encode both "knowledge" and "reasoning" in their parameters. However, such models cannot perform effectively for tasks that require knowledge grounding and tasks that deal with non-stationary data, such as news and social media. Besides, these models often require huge number of parameters to encode all the required knowledge. These issues can be addressed via augmentation with a retrieval model. This category of machine learning models, which is called Retrieval-enhanced machine learning (REML), has recently attracted considerable attention in multiple research communities. For instance, REML models have been studied in the context of open-domain question answering, fact verification, and dialogue systems and also in the context of generalization through memorization in language models and memory networks. We believe that the information retrieval community can significantly contribute to this growing research area by designing, implementing, analyzing, and evaluating various aspects of retrieval models with applications to REML tasks. The goal of this full-day hybrid workshop is to bring together researchers from industry and academia to discuss various aspects of retrieval-enhanced machine learning, including effectiveness, efficiency, and robustness of these models in addition to their impact on real-world applications.

As new forms of data capture emerge to power new AI applications, questions abound about the ethical implications of these data collection practices. In this paper, we present clinicians' perspectives on the prospective benefits and harms of voice data collection during health consultations. Such data collection is being proposed as a means to power models to assist clinicians with medical data entry, administrative tasks, and consultation analysis. Yet, clinicians' attitudes and concerns are largely absent from the AI narratives surrounding these use cases, and the academic literature investigating them. Our qualitative interview study used the concept of an informed consent process as a type of design fiction, to support elicitation of clinicians' perspectives on voice data collection and use associated with a fictional, near-term AI assistant. Through reflexive thematic analysis of in-depth sessions with physicians, we distilled eight classes of potential risks that clinicians are concerned about, including workflow disruptions, self-censorship, and errors that could impact patient eligibility for services. We conclude with an in-depth discussion of these prospective risks, reflect on the use of the speculative processes that illuminated them, and reconsider evaluation criteria for AI-assisted clinical documentation technologies in light of our findings.

Nowadays, most online services are hosted on multi-stakeholder marketplaces, where consumers and producers may have different objectives. Conventional recommendation systems, however, mainly focus on maximizing consumers’ satisfaction by recommending the most relevant items to each individual. This may result in unfair exposure of items, thus jeopardizing producer benefits. Additionally, they do not care whether consumers from diverse demographic groups are equally satisfied. To address these limitations, we propose a multi-objective optimization framework for fairness-aware recommendation,Multi-FR, that adaptively balances accuracy and fairness for various stakeholders with Pareto optimality guarantee. We first propose four fairness constraints on consumers and producers. In order to train the whole framework in an end-to-end way, we utilize the smooth rank and stochastic ranking policy to make these fairness criteria differentiable and friendly to back-propagation. Then, we adopt the multiple gradient descent algorithm to generate a Pareto set of solutions, from which the most appropriate one is selected by the Least Misery Strategy. The experimental results demonstrate thatMulti-FRlargely improves recommendation fairness on multiple stakeholders over the state-of-the-art approaches while maintaining almost the same recommendation accuracy. The training efficiency study confirms our model’s ability to simultaneously optimize different fairness constraints for many stakeholders efficiently.

Recommender systems have become the dominant means of curating cultural content, significantly influencing the nature of individual cultural experience. While the majority of academic and industrial research on recommender systems optimizes for personalized user experience, this paradigm does not capture the ways that recommender systems impact cultural experience in the aggregate, across populations of users. Although existing novelty, diversity, and fairness studies probe how recommender systems relate to the broader social role of cultural content, they do not adequately center culture as a core concept and challenge. In this work, we introduce commonality as a new measure of recommender systems that reflects the degree to which recommendations familiarize a given user population with specified categories of cultural content. Our proposed commonality metric responds to a set of arguments developed through an interdisciplinary dialogue between researchers in computer science and the social sciences and humanities. With reference to principles underpinning non-profit, public service media (PSM) 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. Taking diversity in movie recommendation as a case study in enhancing pluralistic cultural experience, we empirically compare the performance of recommendation algorithms using commonality and existing utility, diversity, novelty, and fairness metrics. Our results demonstrate that commonality captures a property of system behavior complementary to existing metrics and suggest the need for alternative, non-personalized interventions in recommender systems oriented to strengthening cultural citizenship across populations of users. In this way, commonality contributes to a growing body of scholarship developing ‘public good’ rationales for digital media and machine learning systems.

Offline evaluation of information retrieval and recommendation has traditionally focused on distilling the quality of a ranking into a scalar metric such as average precision or normalized discounted cumulative gain. We can use this metric to compare the performance of multiple systems for the same request. Although evaluation metrics provide a convenient summary of system performance, they also collapse subtle differences across users into a single number and can carry assumptions about user behavior and utility not supported across retrieval scenarios. We propose recall-paired preference (RPP), a metric-free evaluation method based on directly computing a preference between ranked lists. RPP simulates multiple user subpopulations per query and compares systems across these pseudo-populations. Our results across multiple search and recommendation tasks demonstrate that RPP substantially improves discriminative power while correlating well with existing metrics and being equally robust to incomplete data.

Natural interaction with recommendation and personalized search systems has received tremendous attention in recent years. We focus on the challenge of supporting people's understanding and control of these systems and explore a fundamentally new way of thinking about representation of knowledge in recommendation and personalization systems. Specifically, we argue that it may be both desirable and possible for algorithms that use natural language representations of users' preferences to be developed. We make the case that this could provide significantly greater transparency, as well as affordances for practical actionable interrogation of, and control over, recommendations. Moreover, we argue that such an approach, if successfully applied, may enable a major step towards systems that rely less on noisy implicit observations while increasing portability of knowledge of one's interests.

Prior research on exposure fairness in the context of recommender systems has focused mostly on disparities in the exposure of individual or groups of items to individual users of the system. The problem of how individual or groups of items may be systemically under or over exposed to groups of users, or even all users, has received relatively less attention. However, such systemic disparities in information exposure can result in observable social harms, such as withholding economic opportunities from historically marginalized groups (allocative harm) or amplifying gendered and racialized stereotypes (representational harm). Previously, Diaz et al. developed the expected exposure metric---that incorporates existing user browsing models that have previously been developed for information retrieval---to study fairness of content exposure to individual users. We extend their proposed framework to formalize a family of exposure fairness metrics that model the problem jointly from the perspective of both the consumers and producers. Specifically, we consider group attributes for both types of stakeholders to identify and mitigate fairness concerns that go beyond individual users and items towards more systemic biases in recommendation. Furthermore, we study and discuss the relationships between the different exposure fairness dimensions proposed in this paper, as well as demonstrate how stochastic ranking policies can be optimized towards said fairness goals.

Although information access systems have long supportedpeople in accomplishing a wide range of tasks, we propose broadening the scope of users of information access systems to include task-driven machines, such as machine learning models. In this way, the core principles of indexing, representation, retrieval, and ranking can be applied and extended to substantially improve model generalization, scalability, robustness, and interpretability. We describe a generic retrieval-enhanced machine learning (REML) framework, which includes a number of existing models as special cases. REML challenges information retrieval conventions, presenting opportunities for novel advances in core areas, including optimization. The REML research agenda lays a foundation for a new style of information access research and paves a path towards advancing machine learning and artificial intelligence.

Search systems control the exposure of ranked content to searchers. In many cases, creators value not only the exposure of their content but, moreover, an understanding of the specific searches where the content is surfaced. The problem of identifying which queries expose a given piece of content in the ranked results is an important and relatively underexplored search transparency challenge. Exposing queries are useful for quantifying various issues of search bias, privacy, data protection, security, and search engine optimization. Exact identification of exposing queries in a given system is computationally expensive, especially in dynamic contexts such as web search. We explore the feasibility of approximate exposing query identification (EQI) as a retrieval task by reversing the role of queries and documents in two classes of search systems: dense dual-encoder models and traditional BM25. We then improve upon this approach through metric learning over the retrieval embedding space. We further derive an evaluation metric to measure the quality of a ranking of exposing queries, as well as conducting an empirical analysis of various practical aspects of approximate EQI. Overall, our work contributes a novel conception of transparency in search systems and computational means of achieving it.

In email interfaces, providing users with reply suggestions may simplify or accelerate correspondence. While the “success” of such systems is typically quantified using the number of suggestions selected by users, this ignores the impact of social context, which can change how suggestions are perceived. To address this, we developed a mixed-methods framework involving qualitative interviews and crowdsourced experiments to characterize problematic email reply suggestions. Our interviews revealed issues with over-positive, dissonant, cultural, and gender-assuming replies, as well as contextual politeness. In our experiments, crowdworkers assessed email scenarios that we generated and systematically controlled, showing that contextual factors like social ties and the presence of salutations impacts users’ perceptions of email correspondence. These assessments created a novel dataset of human-authored corrections for problematic email replies. Our study highlights the social complexity of providing suggestions for email correspondence, raising issues that may apply to all social messaging systems.

While current information retrieval systems are effective for known-item retrieval where the searcher provides a precise name or identifier for the item being sought, systems tend to be much less effective for cases where the searcher is unable to express a precise name or identifier. We refer to this as tip of the tongue (TOT) known-item retrieval, named after the cognitive state of not being able to retrieve an item from memory. Using movie search as a case study, we explore the characteristics of questions posed by searchers in TOT states in a community question answering website. We analyze how searchers express their information needs during TOT states in the movie domain. Specifically, what information do searchers remember about the item being sought and how do they convey this information? Our results suggest that searchers use a combination of information about: (1) the content of the item sought, (2) the context in which they previously engaged with the item, and (3) previous attempts to find the item using other resources (e.g., search engines). Additionally, searchers convey information by sometimes expressing uncertainty (i.e., hedging), opinions, emotions, and by performing relative (vs. absolute) comparisons with attributes of the item. As a result of our analysis, we believe that searchers in TOT states may require specialized query understanding methods or document representations. Finally, our preliminary retrieval experiments show the impact of each information type presented in information requests on retrieval performance.

Designers of online search and recommendation services often need to develop metrics to assess system performance. This tutorial focuses on mixed methods approaches to developing user-focused evaluation metrics. This starts with choosing how data is logged or how to interpret current logged data, with a discussion of how qualitative insights and design decisions can restrict or enable certain types of logging. When we create a metric from that logged data, there are underlying assumptions about how users interact with the system and evaluate those interactions. We will cover what these assumptions look like for some traditional system evaluation metrics and highlight quantitative and qualitative methods that investigate and adapt these assumptions to be more explicit and expressive of genuine user behavior. We discuss the role that mixed methods teams can play at each stage of metric development, starting with data collection, designing both online and offline metrics, and supervising metric selection for decision making. We describe case studies and examples of these methods applied in the context of evaluating personalized search and recommendation systems. Finally, we close with practical advice for applied quantitative researchers who may be in the early stages of planning collaborations with qualitative researchers for mixed methods metrics development.

There is increasing attention to evaluating the fairness of search system ranking decisions. These metrics often consider the membership of items to particular groups, often identified using protected attributes such as gender or ethnicity. To date, these metrics typically assume the availability and completeness of protected attribute labels of items. However, the protected attributes of individuals are rarely present, limiting the application of fair ranking metrics in large scale systems. In order to address this problem, we propose a sampling strategy and estimation technique for four fair ranking metrics. We formulate a robust and unbiased estimator which can operate even with very limited number of labeled items. We evaluate our approach using both simulated and real world data. Our experimental results demonstrate that our method can estimate this family of fair ranking metrics and provides a robust, reliable alternative to exhaustive or random data annotation.

We introduce the concept of expected exposure as the average attention ranked items receive from users over repeated samples of the same query. Furthermore, we advocate for the adoption of the principle of equal expected exposure: given a fixed information need, no item should receive more or less expected exposure than any other item of the same relevance grade. We argue that this principle is desirable for many retrieval objectives and scenarios, including topical diversity and fair ranking. %Leveraging user models from existing retrieval metrics, we propose a general evaluation methodology based on expected exposure and draw connections to related metrics in information retrieval evaluation. Importantly, this methodology relaxes classic information retrieval assumptions, allowing a system, in response to a query, to produce a distribution over rankings instead of a single fixed ranking. We study the behavior of the expected exposure metric and stochastic rankers across a variety of information access conditions, including ad hoc retrieval and recommendation. %We believe that measuring and optimizing expected exposure metrics using randomization opens a new area for retrieval algorithm development and progress.

Article 5(1)(c) of the European Union's General Data Protection Regulation (GDPR) requires that "personal data shall be [...] adequate, relevant, and limited to what is necessary in relation to the purposes for which they are processed (`data minimisation')". To date, the legal and computational definitions of 'purpose limitation' and 'data minimization' remain largely unclear. In particular, the interpretation of these principles is an open issue for information access systems that optimize for user experience through personalization and do not strictly require personal data collection for the delivery of basic service.

Asking clarifying questions in response to search queries has been recognized as a useful technique for revealing the underlying intent of the query. Clarification has applications in retrieval systems with different interfaces, from the traditional web search interfaces to the limited bandwidth interfaces as in speech-only and small screen devices. Generation and evaluation of clarifying questions have been recently studied in the literature. However, user interaction with clarifying questions is relatively unexplored. In this paper, we conduct a comprehensive study by analyzing large-scale user interactions with clarifying questions in a major web search engine. In more detail, we analyze the user engagements received by clarifying questions based on different properties of search queries, clarifying questions, and their candidate answers. We further study click bias in the data, and show that even though reading clarifying questions and candidate answers does not take significant efforts, there still exist some position and presentation biases in the data. We also propose a model for learning representation for clarifying questions based on the user interaction data as implicit feedback. The model is used for re-ranking a number of automatically generated clarifying questions for a given query. Evaluation on both click data and human labeled data demonstrates the high quality of the proposed method.

Problematic web search query completion suggestions-perceived as biased, offensive, or in some other way harmful-can reinforce existing stereotypes and misbeliefs, and even nudge users towards undesirable patterns of behavior. Locating such suggestions is difficult, not only due to the long-tailed nature of web search, but also due to differences in how people assess potential harms. Grounding our study in web search query logs, we explore when system-provided suggestions might be perceived as problematic through a series of crowd-experiments where we systematically manipulate: the search query fragments provided by users, possible user search intents, and the list of query completion suggestions. To examine why query suggestions might be perceived as problematic, we contrast them to an inventory of known types of problematic suggestions. We report our observations around differences in the prevalence of a) suggestions that are problematic on their own versus b) suggestions that are problematic for the query fragment provided by a user, for both common informational needs and in the presence of web search voids-topics searched by few to no users. Our experiments surface a rich array of scenarios where suggestions are considered problematic, including due to the context in which they were surfaced. Compounded by the elusive nature of many such scenarios, the prevalence of suggestions perceived as problematic only for certain user inputs, raises concerns about blind spots due to data annotation practices that may lead to some types of problematic suggestions being overlooked.

Fairness and related concerns have become of increasing importance in a variety of AI and machine learning contexts. They are also highly relevant to recommender systems and related problems such as information retrieval, as evidenced by the growing literature in RecSys, FAT*, SIGIR, and special sessions such as the FATREC and FACTS-IR workshops and the Fairness track at TREC 2019; however, translating algorithmic fairness constructs from classification, scoring, and even many ranking settings into recommendation and other information access scenarios is not a straightforward task. This tutorial will help orient RecSys researchers to algorithmic fairness, understand how concepts do and do not translate from other settings, and provide an introduction to the growing literature on this topic.

Fairness and related concerns have become of increasing importance in a variety of AI and machine learning contexts. They are also highly relevant to information retrieval and related problems such as recommendation, as evidenced by the growing literature in SIGIR, FAT*, RecSys, and special sessions such as the FATREC workshop and the Fairness track at TREC 2019; however, translating algorithmic fairness constructs from classification, scoring, and even many ranking settings into information retrieval and recommendation scenarios is not a straightforward task. This tutorial will help to orient IR researchers to algorithmic fairness, understand how concepts do and do not translate from other settings, and provide an introduction to the growing literature on this topic.

Two-sided marketplaces are platforms that have customers not only on the demand side (e.g. users), but also on the supply side (e.g. retailer, artists). While traditional recommender systems focused specifically towards increasing consumer satisfaction by providing relevant content to consumers, two-sided marketplaces face the problem of additionally optimizing for supplier preferences, and visibility. Indeed, the suppliers would want afair opportunity to be presented to users. Blindly optimizing for consumer relevance may have a detrimental impact on supplier fairness. Motivated by this problem, we focus on the trade-off between objectives of consumers and suppliers in the case of music streaming services, and consider the trade-off betweenrelevance of recommendations to the consumer (i.e. user) andfairness of representation of suppliers (i.e. artists) and measure their impact on consumersatisfaction.

Evaluation is a fundamental part of a recommendation system. Evaluation typically takes one of three forms: (1) smaller lab studies with real users; (2) batch tests with offline collections, judgements, and measures; (3) large-scale controlled experiments (e.g. A/B tests) looking at implicit feedback. But it is rare for the first to inform and influence the latter two; in particular, implicit feedback metrics often have to be continuously revised and updated as assumptions are found to be poorly supported.

We study the use and evaluation of a system for supporting music discovery, the experience of finding and listening to content previously unknown to the user. We adopt a mixed methods approach, including interviews, unsupervised learning, survey research, and statistical modeling, to understand and evaluate user satisfaction in the context of discovery. User interviews and survey data show that users' behaviors change according to their goals, such as listening to recommended tracks in the moment, or using recommendations as a starting point for exploration. We use these findings to develop a statistical model of user satisfaction at scale from interactions with a music streaming platform. We show that capturing users' goals, their deviations from their usual behavior, and their peak interactions on individual tracks are informative for estimating user satisfaction. Finally, we present and validate heuristic metrics that are grounded in user experience for online evaluation of recommendation performance. Our findings, supported with evidence from both qualitative and quantitative studies, reveal new insights about user expectations with discovery and their behavioral responses to satisfying and dissatisfying systems.

In recent years, machine learning approaches, and in particular deep neural networks, have yielded significant improvements on several natural language processing and computer vision tasks; however, such breakthroughs have not yet been observed in the area of information retrieval. Besides the complexity of IR tasks, such as understanding the user's information needs, a main reason is the lack of high-quality and/or large-scale training data for many IR tasks. This necessitates studying how to design and train machine learning algorithms where there is no large-scale or high-quality data in hand. Therefore, considering the quick progress in development of machine learning models, this is an ideal time for a workshop that especially focuses on learning in such an important and challenging setting for IR tasks. The goal of this workshop is to bring together researchers from industry---where data is plentiful but noisy---with researchers from academia---where data is sparse but clean to discuss solutions to these related problems.

Spotify provides users with access to a massive repository of streaming music. While some aspects of music access are familiar to the information retrieval community (e.g. semistructured data, item recommendation), nuances of the music domain require the development of new models of user understanding, intent modeling, relevance, and content understanding. These models can be studied using the large amount of content and usage data at Spotify, allowing us to extend previous results in the music information retrieval community. In this presentation, we will highlight the research involved in developing Spotify and outline a research program for large scale music access.

In recent years, the information retrieval (IR) community has witnessed the first successful applications of deep neural network models to short-text matching and ad-hoc retrieval tasks. However, the two communities - focused on deep neural networks and on IR - have less in common when it comes to the choice of programming languages. Indri, an indexing framework popularly used by the IR community, is written in C++, while Torch, a popular machine learning library for deep learning, is written in the light-weight scripting language Lua. To bridge this gap, we introduce Luandri (pronounced "laundry"), a simple interface for exposing the search capabilities of Indri to Torch models implemented in Lua.

Models such as latent semantic analysis and those based on neural embeddings learn distributed representations of text, and match the query against the document in the latent semantic space. In traditional information retrieval models, on the other hand, terms have discrete or local representations, and the relevance of a document is determined by the exact matches of query terms in the body text. We hypothesize that matching with distributed representations complements matching with traditional local representations, and that a combination of the two is favourable. We propose a novel document ranking model composed of two separate deep neural networks, one that matches the query and the document using a local representation, and another that matches the query and the document using learned distributed representations. The two networks are jointly trained as part of a single neural network. We show that this combination or 'duet' performs significantly better than either neural network individually on a Web page ranking task, and significantly outperforms traditional baselines and other recently proposed models based on neural networks.

Search result examination is an important part of searching. High page load latency for landing pages (clicked search results) can reduce the efficiency of the search process. Proactively prefetching landing pages in advance of clickthrough can save searchers valuable time. However, prefetching consumes resources (primarily bandwidth and battery) that are wasted unless the prefetched results are requested by searchers. Balancing the costs in prefetching particular results against the benefits in reduced latency to searchers represents the search result prefetching challenge. In this article, we introduce this challenge and present methods to address it in both desktop and mobile settings. Our methods leverage searchers’ cursor movements (on desktop) and viewport-based viewing behavior (on mobile) on search engine result pages (SERPs) in real time to dynamically estimate the result that searchers will request next. We demonstrate through large-scale log analysis that our approach significantly outperforms three strong baselines that prefetch results based on (i) the search engine result ranking (prefetch top-ranked results), (ii) past SERP clicks from all searchers for the query (prefetch popular results), or (iii) past SERP clicks from the current searcher for the query (prefetch results that the searcher prefers). Our promising findings have implications for the design of search support in desktop and mobile settings that makes the search process more efficient.

Continuous space word embeddings have received a great deal of attention in the natural language processing and machine learning communities for their ability to model term similarity and other relationships.We study the use of term relatedness in the context of query expansion for ad hoc information retrieval.We demonstrate that word embeddings such as word2vec and GloVe, when trained globally, underperform corpus and query specific embeddings for retrieval tasks.These results suggest that other tasks benefiting from global embeddings may also benefit from local embeddings.

The proliferation of social media platforms together with the wide adoption of smartphone devices has transformed how we communicate and share news. During large-scale emergencies, such as natural disasters or armed attacks, victims, responders, and volunteers increasingly use social media to post situation updates and to request and offer help. The use of social media for emergency and disaster response has been a prominent application of information and knowledge management techniques in recent years. There are a number of challenges associated with near real-time processing of vast volumes of information in a way that makes sense for people directly affected, for volunteer organizations, and for official emergency response agencies. As massive amount of messages posted by users are transformed into semi-structured records via information extraction and natural language processing techniques, there is a growing need for developing advanced techniques to aggregate this large-scale data to gain an understanding of the ``big picture'' of an emergency, and to detect and predict how a disaster could develop. This workshop seeks to provide a platform for the exchange of ideas, identification of important problems, and discovery of possible synergies. It will enable interesting discussions and encouraged collaboration between various disciplines, and information and knowledge management approaches is the core of this workshop.

Unexpected news events, such as natural disasters or other human tragedies, create a large volume of dynamic text data from official news media as well as less formal social media. Automatic real-time text summarization has become an important tool for quickly transforming this overabundance of text into clear, useful information for end-users including affected individuals, crisis responders, and interested third parties. Despite the importance of real-time summarization systems, their evaluation is not well understood as classic methods for text summarization are inappropriate for real-time and streaming conditions.

Search result examination is an important part of searching. High page load latency for landing pages (clicked results) can reduce the efficiency of the search process. Proactively prefetching landing pages in advance of clickthrough can save searchers valuable time. However, prefetching consumes resources that are wasted unless the prefetched results are requested by searchers. Balancing the costs in prefetching particular results against the benefits in reduced latency to searchers represents the search result prefetching challenge. We present methods that leverage searchers' cursor movements on search result pages in real time to dynamically estimate the result that searchers will request next. We demonstrate through large-scale log analysis that our approach significantly outperforms three strong baselines that prefetch results based on (i) the search engine result ranking, (ii) past clicks from all searchers for the query, or (iii) past clicks from the current searcher for the query. Our promising findings have implications for the design of search support that makes the search process more efficient.

No abstract available.

Chris Kedzie, Kathleen McKeown, Fernando Diaz. Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2015.

No abstract available.

We study the trade-off between layout elements of the search results page and revenue in the real-time sponsored search auction. Using data from a randomized experiment on a major search engine, we find that having images present among the search results tends to simultaneously raise the ad click-through rate and flatten the ad click curve, reducing the premium for occupying the top slot and thus impacting bidding incentives. Theoretical analysis shows that this type of change creates an ambiguous impact on revenue in equilibrium: a steeper curve with lower total click-through rate is preferable only if the expected revenue distribution is skewed enough towards the top bidder. Empirically, we show that this is a relatively rare phenomenon, and we also find that whole page satisfaction causally raises the click-through rate of the ad block. This means search engines have a short-run incentive to boost search result quality, not just a long-run incentive based on competition between providers.

No abstract available.

Users frequently interact with web search systems on their mobile devices via multiple modalities, including touch and speech. These interaction modes are substantially different from the user experience on desktop search. As a result, system designers have new challenges and questions around understanding the intent on these platforms. In this paper, we study the query reformulation patterns in mobile logs. We group query reformulations based on their input method into four categories; text-text, text-voice, voice-text and voice-voice. We discuss the unique characteristics of each of these groups by comparing them against each other and desktop logs. We also compare the distribution of reformulation types (e.g. adding/dropping words) against desktop logs and show that there are new classes of reformulations that are caused by errors in speech recognition. Our results suggest that users do not tend to switch between different input types (e.g. voice and text). Voice to text switches are largely caused by speech recognition errors, and text to voice switches are unlikely to be about the same intent.

Document-level relevance judgments are a major component in the calculation of effectiveness metrics. Collecting high-quality judgments is therefore a critical step in information retrieval evaluation. However, the nature of and the assumptions underlying relevance judgment collection have not received much attention. In particular, relevance judgments are typically collected for each document in isolation, although users read each document in the context of other documents. In this work, we aim to investigate the nature of relevance judgment collection. We collect relevance labels in both isolated and conditional setting, and ask for judgments in various dimensions of relevance as well as overall relevance. Then we compare the relevance metrics based on various types of judgments with other metrics of quality such as user preference. Our analyses illuminate how these settings for judgment collection affect the quality and the characteristics of the judgments. We also find that the metrics based on conditional judgments show higher correlation with user preference than isolated judgments.

Understanding how users examine result pages across a broad range of information needs is critical for search engine design. Cursor movements can be used to estimate visual attention on search engine results page (SERP) components, including traditional snippets, aggregated results, and advertisements. However, these signals can only be leveraged for SERPs where cursor tracking was enabled, limiting their utility for informing the design of new SERPs. In this work, we develop robust, log-based mouse movement models capable of estimating searcher attention on novel SERP arrangements. These models can help improve SERP design by anticipating searchers' engagement patterns given a proposed arrangement. We demonstrate the efficacy of our method using a large set of mouse-tracking data collected from two independent commercial search engines.

Web content increasingly reflects the current state of the physical and social world, manifested both in traditional news media sources along with user-generated publishing sites such as Twitter, Foursquare, and Facebook. At the same time, web searching increasingly reflects problems grounded in the real world. As a result of this blending of the web with the real world, we observe that the web, both in its composition and use, has incorporated many of the dynamics of the real world. Few of the problems associated with searching dynamic collections are well understood, such as defining time-sensitive relevance, understanding user query behavior over time and understanding why certain web content changes.

The World Wide Web is highly dynamic and is constantly evolving to cover the latest information about the physical and social updates in the world. At the same time, the changes in web contents are entangled with new information needs and time-sensitive user interactions with information sources. To address these temporal information needs effectively, it is essential for the search engines to model web dynamics and understand the changes in user behavior over time that are caused by them.

In Web search and vertical search, recency ranking refers to retrieving and ranking documents by both relevance and freshness. As impoverished in-links and click information is the the biggest challenge for recency ranking, we advocate the use of Twitter data to address the challenge in this article. We propose a method to utilize Twitter TinyURL to detect fresh and high-quality documents, and leverage Twitter data to generate novel and effective features for ranking. The empirical experiments demonstrate that the proposed approach effectively improves a commercial search engine for both Web search ranking and tweet vertical ranking.

The information need of users and the documents which answer this need are frequently contingent on the different characteristics of users. This is especially evident during natural disasters, such as earthquakes and violent weather incidents, which create a strong transient information need. In this article, we investigate how the information need of users, as expressed by their queries, is affected by their physical detachment, as estimated by their physical location in relation to that of the event, and by their social detachment, as quantified by the number of their acquaintances who may be affected by the event. Drawing on large-scale data from ten major events, we show that social and physical detachment levels of users are a major influence on their search engine queries. We demonstrate how knowing social and physical detachment levels can assist in improving retrieval for two applications: identifying search queries related to events and ranking results in response to event-related queries. We find that the average precision in identifying relevant search queries improves by approximately 18%, and that the average precision of ranking that uses detachment information improves by 10%. Using both types of detachment achieved a larger gain in performance than each of them separately.

Aggregated search is the task of integrating results from potentially multiple specialized search services, or verticals, into the Web search results. The task requires predicting not only which verticals to present (the focus of most prior research), but also predicting where in the Web results to present them (i.e., above or below the Web results, or somewhere in between). Learning models to aggregate results from multiple verticals is associated with two major challenges. First, because verticals retrieve different types of results and address different search tasks, results from different verticals are associated with different types of predictive evidence (or features). Second, even when a feature is common across verticals, its predictiveness may be vertical-specific. Therefore, approaches to aggregating vertical results require handling an inconsistent feature representation across verticals, and, potentially, a vertical-specific relationship between features and relevance. We present 3 general approaches that address these challenges in different ways and compare their results across a set of 13 verticals and 1070 queries. We show that the best approaches are those that allow the learning algorithm to learn a vertical-specific relationship between features and relevance.

The information needs of users and the documents which answer it are frequently contingent on the different characteristics of users. This is especially evident during natural disasters, such as earthquakes and violent weather incidents, which create a strong transient information need. In this paper we investigate how the information need of users is affected by their physical detachment, as estimated by their physical location in relation to that of the event, and by their social detachment, as quantified by the number of their acquaintances who may be affected by the event. Drawing on large-scale data from three major events, we show that social and physical detachment levels of users are a major influence on their information needs, as manifested by their search engine queries. We demonstrate how knowing social and physical detachment levels can assist in improving retrieval for two applications: identifying search queries related to events and ranking results in response to event-related queries. We find that the average precision in identifying relevant search queries improves by approximately 18%, and that the average precision of ranking that uses detachment information improves by 10%.

People nowadays can obtain information on current news events through media outlets, social media, and by actively seeking information using search engines. In this paper we investigate the temporal relationship between news coverage by media outlets, social media, and query logs and show that social media frequently precedes other information sources. Additionally, we demonstrate that there is strong negative correlation between the probability for reporting of an event and the distance of the information source from the event.

Proactive link suggestion leads to improved user experience by allowing users to reach relevant information with fewer clicks, fewer pages to read, or simply faster because the right pages are prefetched just in time. In this paper we tackle two new scenarios for link suggestion, which were not covered in prior work owing to scarcity of historical browsing data. In the web search scenario, we propose a method for generating quick links - additional entry points into Web sites, which are shown for top search results for navigational queries - for tail sites, for which little browsing statistics is available. Beyond Web search, we also propose a method for link suggestion in general web browsing, effectively anticipating the next link to be followed by the user. Our approach performs clustering of Web sites in order to aggregate information across multiple sites, and enables relevant link suggestion for virtually any site, including tail sites and brand new sites for which little historical data is available. Empirical evaluation confirms the validity of our method using editorially labeled data as well as real-life search and browsing data from a major US search engine.

Vertical aggregation is the task of incorporating results from specialized search engines or verticals (e.g., images, video, news) into Web search results. Vertical selection is the subtask of deciding, given a query, which verticals, if any, are relevant. State of the art approaches use machine learned models to predict which verticals are relevant to a query. When trained using a large set of labeled data, a machine learned vertical selection model outperforms baselines which require no training data. Unfortunately, whenever a new vertical is introduced, a costly new set of editorial data must be gathered. In this paper, we propose methods for reusing training data from a set of existing (source) verticals to learn a predictive model for a new (target) vertical. We study methods for learning robust, portable, and adaptive cross-vertical models. Experiments show the need to focus on different types of features when maximizing portability (the ability for a single model to make accurate predictions across multiple verticals) than when maximizing adaptability (the ability for a single model to make accurate predictions for a specific vertical). We demonstrate the efficacy of our methods through extensive experimentation for 11 verticals

Federated search refers to the brokered retrieval of content from a set of auxiliary retrieval systems instead of from a single, centralized retrieval system. Federated search tasks occur in, for example, digital libraries (where documents from several retrieval systems must be seamlessly merged) or peer-to-peer information retrieval (where documents distributed across a network of local indexes must be retrieved).

Match-making systems refer to systems where users want to meet other individuals to satisfy some underlying need. Examples of match-making systems include dating services, resume/job bulletin boards, community based question answering, and consumer-to-consumer marketplaces. One fundamental component of a match-making system is the retrieval and ranking of candidate matches for a given user. We present the first in-depth study of information retrieval approaches applied to match-making systems. Specifically, we focus on retrieval for a dating service. This domain offers several unique problems not found in traditional information retrieval tasks. These include two-sided relevance, very subjective relevance, extremely few relevant matches, and structured queries. We propose a machine learned ranking function that makes use of features extracted from the uniquely rich user profiles that consist of both structured and unstructured attributes. An extensive evaluation carried out using data gathered from a real online dating service shows the benefits of our proposed methodology with respect to traditional match-making baseline systems. Our analysis also provides deep insights into the aspects of match-making that are particularly important for producing highly relevant matches.

In web search, recency ranking refers to ranking documents by relevance which takes freshness into account. In this paper, we propose a retrieval system which automatically detects and responds to recency sensitive queries. The system detects recency sensitive queries using a high precision classifier. The system responds to recency sensitive queries by using a machine learned ranking model trained for such queries. We use multiple recency features to provide temporal evidence which effectively represents document recency. Furthermore, we propose several training methodologies important for training recency sensitive rankers. Finally, we develop new evaluation metrics for recency sensitive queries. Our experiments demonstrate the efficacy of the proposed approaches.

Realtime web search refers to the retrieval of very fresh content which is in high demand. An effective portal web search engine must support a variety of search needs, including realtime web search. However, supporting realtime web search introduces two challenges not encountered in non-realtime web search: quickly crawling relevant content and ranking documents with impoverished link and click information. In this paper, we advocate the use of realtime micro-blogging data for addressing both of these problems. We propose a method to use the micro-blogging data stream to detect fresh URLs. We also use micro-blogging data to compute novel and effective features for ranking fresh URLs. We demonstrate these methods improve effective of the portal web search engine for realtime web search.

In some retrieval situations, a system must search across multiple collections. This task, referred to as federated search, occurs for example when searching a distributed index or aggregating content for web search. Resource selection refers to the subtask of deciding, given a query, which collections to search. Most existing resource selection methods rely on evidence found in collection content. We present an approach to resource selection that combines multiple sources of evidence to inform the selection decision. We derive evidence from three different sources: collection documents, the topic of the query, and query click-through data. We combine this evidence by treating resource selection as a multiclass machine learning problem. Although machine learned approaches often require large amounts of manually generated training data, we present a method for using automatically generated training data. We make use of and compare against prior resource selection work and evaluate across three experimental testbeds.

Geographic information retrieval encompasses important tasks including finding the location of a user, and locations relevant to their search queries. Web-based search engines receive queries from numerous users located in very different parts of the world. A typical way for people to find news is through a general web search engine, which makes it important for search engines to recognize queries with news intent. An important question for geographic information retrieval is how we can benefit from geographic cues to predict the intent of users. This work presents a case study of an application using geographic features to improve the quality of an important web search task, involving predicting which queries have news intent and hence are likely to receive clicks on news search results. Our case study suggests that information derived from geographic features can help the task. The information we consider includes cues derived from the location of the user, from the IP address, the location relevant to the query, automatically extracted from the query string, and the relation between the two locations. We build a classifier that uses geographical cues to predict whether a query will result in a news click or not. We compare our classifier to a strong baseline that use non-geographic click-based features and we show that our classifier outperforms the baseline for geographic queries.

Web search providers often include search services for domain-specific subcollections, called verticals, such as news, images, videos, job postings, company summaries, and artist profiles. We address the problem of vertical selection, predicting relevant verticals (if any) for queries issued to the search engine's main web search page. In contrast to prior query classification and resource selection tasks, vertical selection is associated with unique resources that can inform the classification decision. We focus on three sources of evidence: (1) the query string, from which features are derived independent of external resources, (2) logs of queries previously issued directly to the vertical, and (3) corpora representative of vertical content. We focus on 18 different verticals, which differ in terms of semantics, media type, size, and level of query traffic. We compare our method to prior work in federated search and retrieval effectiveness prediction. An in-depth error analysis reveals unique challenges across different verticals and provides insight into vertical selection for future work.

Web search results often integrate content from specialized corpora known as verticals. Given a query, one important aspect of aggregated search is the selection of relevant verticals from a set of candidate verticals. One drawback to previous approaches to vertical selection is that methods have not explicitly modeled user feedback. However, production search systems often record a variety of feedback information. In this paper, we present algorithms for vertical selection which adapt to user feedback. We evaluate algorithms using a novel simulator which models performance of a vertical selector situated in realistic query traffic.

Aggregated search refers to the integration of content from specialized corpora or verticals into web search results. Aggregation improves search when the user has vertical intent but may not be aware of or desire vertical search. In this paper, we address the issue of integrating search results from a news vertical into web search results. News is particularly challenging because, given a query, the appropriate decision---to integrate news content or not---changes with time. Our system adapts to news intent in two ways. First, by inspecting the dynamics of the news collection and query volume, we can track development of and interest in topics. Second, by using click feedback, we can quickly recover from system errors. We define several click-based metrics which allow a system to be monitored and tuned without annotator effort.

We present a method for projecting retrieval scores across two corpora with a shared, parallel corpus.

We present theoretical bounds and empirical robustness of score regularization given changes in the similarity measure.

Evaluation of information retrieval systems is one of the core tasks in information retrieval. Problems include the inability to exhaustively label all documents for a topic, generalizability from a small number of topics, and incorporating the variability of retrieval systems. Previous work addresses the evaluation of systems, the ranking of queries by difficulty, and the ranking of individual retrievals by performance. Approaches exist for the case of few and even no relevance judgments. Our focus is on zero-judgment performance prediction of individual retrievals. One common shortcoming of previous techniques is the assumption of uncorrelated document scores and judgments. If documents are embedded in a high-dimensional space (as they often are), we can apply techniques from spatial data analysis to detect correlations between document scores. We find that the low correlation between scores of topically close documents often implies a poor retrieval performance. When compared to a state of the art baseline, we demonstrate that the spatial analysis of retrieval scores provides significantly better prediction performance. These new predictors can also be incorporated with classic predictors to improve performance further. We also describe the first large-scale experiment to evaluate zero-judgment performance prediction for a massive number of retrieval systems over a variety collections in several languages.

Documents with timestamps, such as email and news, can be placed along a timeline. The timeline for a set of documents returned in response to a query gives an indication of how documents relevant to that query are distributed in time. Examining the timeline of a query result set allows us to characterize both how temporally dependent the topic is, as well as how relevant the results are likely to be. We outline characteristic patterns in query result set timelines, and show experimentally that we can automatically classify documents into these classes. We also show that properties of the query result set timeline can help predict the mean average precision of a query. These results show that meta-features associated with a query can be combined with text retrieval techniques to improve our understanding and treatment of text search on documents with timestamps.

Information retrieval algorithms leverage various collection statistics to improve performance. Because these statistics are often computed on a relatively small evaluation corpus, we believe using larger, non-evaluation corpora should improve performance. Specifically, we advocate incorporating external corpora based on language modeling. We refer to this process as external expansion. When compared to traditional pseudo-relevance feedback techniques, external expansion is more stable across topics and up to 10% more effective in terms of mean average precision. Our results show that using a high quality corpus that is comparable to the evaluation corpus can be as, if not more, effective than using the web. Our results also show that external expansion outperforms simulated relevance feedback. In addition, we propose a method for predicting the extent to which external expansion will improve retrieval performance. Our new measure demonstrates positive correlation with improvements in mean average precision.

The cluster hypothesis states: closely related documents tend to be relevant to the same request. We exploit this hypothesis directly by adjusting ad hoc retrieval scores from an initial retrieval so that topically related documents receive similar scores. We refer to this process as score regularization. Score regularization can be presented as an optimization problem, allowing the use of results from semisupervised learning. We demonstrate that regularized scores consistently and significantly rank documents better than un-regularized scores, given a variety of initial retrieval algorithms. We evaluate our method on two large corpora across a substantial number of topics.

A key missing component in information retrieval systems is self-diagnostic tests to establish whether the system can provide reasonable results for a given query on a document collection. If we can measure properties of a retrieved set of documents which allow us to predict average precision, we can automate the decision of whether to elicit relevance feedback, or modify the retrieval system in other ways. We use meta-data attached to documents in the form of time stamps to measure the distribution of documents retrieved in response to a query, over the time domain, to create a temporal profile for a query. We define some useful features over this temporal profile. We find that using these temporal features, together with the content of the documents retrieved, we can improve the prediction of average precision for a query.