We investigate between-user drivers of query variability through a controlled between-subject, full-factorial user study that manipulates age, gender, and language proficiency across six backstory-driven search tasks. From initial queries, session logs, and post-task interviews, we quantify how demographic and task factors shape query-, task-, and session-level behaviors. We further derive a small set of interpretable latent search dimensions from user evidence to analyze and simulate heterogeneous query behavior. Our results show that age is the most consistent predictor of query formulation and search interaction patterns. Gender and language differences are more selective, and task context is further associated with these patterns. The latent dimensions help explain the differences as variation in search strategy rather than uniform differences in engagement or ability. The paper provides a trait-informed view of heterogeneous search behavior that supports more user-aware analysis and robustness-oriented evaluation in IR.

Large language models (LLMs) are widely used for question answering (QA) but can generate biased or stereotype-driven answers due to demographic associations learned during pre-training. Existing mitigation strategies often rely on model access or fine-tuning, which limits their applicability to closed-source LLMs. We propose a Causal Front Door Prompting framework (CFDP) that reduces demographic influence by intervening on the chain of thought reasoning, which is treated as an observable mediator. CFDP samples and clusters multiple reasoning traces and estimates answer probabilities through weighted aggregation. Experiments on two widely used bias-sensitive QA benchmarks, BBQ and Stereotype, across major LLMs show that CFDP consistently improves fairness metrics without sacrificing QA accuracy. Ablation and sensitivity analyses confirm the value of each component, indicating that causal intervention on reasoning provides an effective and practical approach for bias mitigation in LLM-based QA.

Recommenders built upon implicit collaborative filtering are typically trained to distinguish between users' positive and negative preferences. When direct observations of the latter are unavailable, negative training data are constructed with sampling techniques. But since items often exhibit clustering in the latent space, existing methods tend to oversample negatives from dense regions, resulting in homogeneous training data and limited model expressiveness. To address these shortcomings, we propose a novel negative sampler with diversity guarantees. To achieve them, our approach first pairs each positive item of a user with one that they have not yet interacted with; this instance, called hard negative, is chosen as the top-scoring item according to the model. Instead of discarding the remaining highly informative items, we store them in a user-specific cache. Next, our diversity-augmented sampler selects a representative subset of negatives from the cache, ensuring its dissimilarity from the corresponding user's hard negatives. Our generator then combines these items with the hard negatives, replacing them to produce more effective (synthetic) negative training data that are informative and diverse. Experiments show that our method consistently leads to superior recommendation quality without sacrificing computational efficiency.

Large Language Models (LLMs) are increasingly used to replace human judges to assess the relevance of information objects, raising concerns about circularity, bias, and whether simulated preferences can substitute for human judgement. This work presents experiments using multiple LLMs to label passages for relevance. It examines their gullibility—how easily they are misled into labelling irrelevant passages as relevant. It also compares LLMs with human judges in ranking systems, analysing differences in discriminative power and whether some systems benefit under LLM-based evaluation. Results show that LLMs are influenced by the presence of query terms, even with irrelevant or random passages. Moreover, LLM-generated rankings are highly correlated with those of human judges, with strong agreement on which system is better in pairwise comparisons. However, LLMs may exhibit lower discriminative power, as seen in flatter ranking slopes and missed significance for meaningful improvements. Yet, there are no cases where capable LLMs and human judges reach opposing conclusions with significance. LLMs may boost traditional systems more than neural ones, adding a new concern of system bias. These findings highlight the strong potential of LLMs for relevance labelling, while also highlighting failure cases that call for careful adoption and further research to maintain evaluation integrity. 1

Recommender systems trained on implicit feedback data rely on negative sampling to distinguish positive items from negative items for each user. Since the majority of positive interactions come from a small group of active users, negative samplers are often impacted by data imbalance, leading them to choose more informative negatives for prominent users while providing less useful ones for users who are not so active. This leads to inactive users being further marginalised in the training process, thus receiving inferior recommendations. In this paper, we conduct a comprehensive empirical study demonstrating that state-of-the-art negative sampling strategies provide more accurate recommendations for active users than for inactive users. We also find that increasing the number of negative samples for each positive item improves the average performance, but the benefit is distributed unequally across user groups, with active users experiencing performance gain while inactive users suffering performance degradation. To address this, we propose a group-specific negative sampling strategy that assigns smaller negative ratios to inactive user groups and larger ratios to active groups. Experiments on eight negative samplers show that our approach improves user-side fairness and performance when compared to a uniform global ratio.

Large Language Models (LLMs) are increasingly used to generate search queries for various Information Retrieval (IR) tasks. However, it remains unclear how these machine-generated queries compare to human-written ones, particularly in terms of diversity and alignment with real user behavior. This paper presents an empirical comparison of LLM- and human-generated queries across multiple dimensions, including lexical diversity, linguistic variation, and retrieval effectiveness. We analyze queries produced by several LLMs and compare them with human queries from two datasets collected five years apart. Our findings show that while LLMs can generate diverse queries, their patterns differ from those observed in human behavior. LLM queries typically exhibit higher surface-level uniqueness but rely less on stopword use and word form variation. They also achieve lower retrieval effectiveness when judged against human queries, suggesting that LLM-generated queries may not always reflect real user intent. These differences highlight the limitations of current LLMs in replicating natural querying behavior. We discuss the implications of these findings for LLM-based query generation and user behavior simulation in IR. We conclude that while LLMs hold potential, they should be used with caution.

Formulating queries is a challenging stage of the search process. This study investigates how crowd workers formulate an initial query for a common information need described in a backstory, resulting in diverse query variations. Using the UQV100 dataset of information need backstories and corresponding queries, we analyze the variations. Our findings show that 70% of the query terms used in crowd worker queries did not appear in the backstory text. Examining such terms we developed a taxonomy of search strategies, with the most common being semantic variations of backstory terms, followed by information type specifications. Additionally, we categorized the backstories by cognitive complexity, showing that higher complexity led to greater diversity in query variations and a wider range of term variant categories. This study highlights the importance of accounting for query variations, term variants, user strategies, and cognitive complexity in designing search systems and test collections to better align with users' information needs, influenced by the cognitive demands of a task, and enhance system performance and usability.

Recommender systems play a crucial role in personalizing user experiences, yet ensuring fairness in their outcomes remains an elusive challenge. This work explores the impact of individual users or items on the fairness of recommender systems, thus addressing a significant knowledge gap in the field. We introduce an innovative approach called Adding-Based Counterfactual Fairness Reasoning ( ACFR ), designed to elucidate recommendation fairness from the unique perspectives of users and items. Conventional methodologies, like erasing-based counterfactual analysis, pose limitations, particularly in modern recommender systems dealing with a large number of users and items. These traditional methods, by excluding specific users or items, risk disrupting the crucial relational structure central to collaborative filtering recommendations. In contrast, ACFR employs an adding-based counterfactual analysis, a unique strategy allowing us to consider potential, yet-to-happen user-item interactions. This strategy preserves the core user-item relational structure, while predicting future behaviors of users or items. The commonly used feature-based counterfactual analysis, relying on gradient-based optimization to identify interference on each feature, is not directly applicable in our case. In the recommendation scenario we consider, only interactions between users and items are present during model training—no distinct features are involved. Consequently, the traditional mechanism proves impractical for identifying interference on these existing interactions. Our extensive experiments validate the superiority of ACFR over traditional baseline methods, demonstrating significant improvements in recommendation fairness on benchmark datasets. This work, therefore, provides a fresh perspective and a promising methodology for enhancing fairness in recommender systems.

The effectiveness of clarification question models in engaging users within search systems is currently constrained, casting doubt on their overall usefulness. To improve the performance of these models, it is crucial to employ assessment approaches that encompass both real-time feedback from users (online evaluation) and the characteristics of clarification questions evaluated through human assessment (offline evaluation). However, the relationship between online and offline evaluations has been debated in information retrieval. This study aims to investigate how this discordance holds in search clarification. We use user engagement as ground truth and employ several offline labels to investigate to what extent the offline ranked lists of clarification resemble the ideal ranked lists based on online user engagement. Contrary to the current understanding that offline evaluations fall short of supporting online evaluations, we indicate that when identifying the most engaging clarification questions from the user’s perspective, online and offline evaluations correspond with each other. We show that the query length does not influence the relationship between online and offline evaluations, and reducing uncertainty in online evaluation strengthens this relationship. We illustrate that an engaging clarification needs to excel from multiple perspectives, and SERP quality and characteristics of the clarification are equally important. We also investigate if human labels can enhance the performance of Large Language Models (LLMs) and Learning-to-Rank (LTR) models in identifying the most engaging clarification questions from the user’s perspective by incorporating offline evaluations as input features. Our results indicate that LTR models do not perform better than individual offline labels. However, GPT, an LLM, emerges as the standout performer, surpassing all LTR models and offline labels.

We examine the context of significance tests in offline retrieval experiments. Our Information Retrieval (IR) community is notable for its experimental rigour: the use of statistical significance is grows across our publications. However, we show that ignoring the context of a test risks Type I errors, leading to potential publication bias. We examine two contexts: multiple testing and the types of the retrieval systems being compared. Our results show that multiple testing corrections are critical for experimental work. In addition, we find that past research on the reliability of test collections maybe flawed owing to the type of systems examined. The latter result has not been shown before. Together our results suggest substantial numbers of Type I errors in offline IR experiments. We detail a methodology to alleviate the errors.

Many existing recommender systems (RSs) assume user behavior is governed solely by their interests. However, the peer effect often influences individual decision-making, which leads to conformity behavior. Conventional solutions that eliminate indiscriminately such bias may cause RSs to neglect valuable information and depersonalize the recommendation results. Also, conformity can transform into user interest, e.g., discovering new tastes after a glance at popular music. By better representing different forms of conformity influence, we can do a better job at interest mining and debiasing. In certain extreme circumstances, the herd effect may be exacerbated by user anxiety with uncertainty (e.g., panic buying during the COVID-19 pandemic). RSs may thus fail to respond in time due to sudden and dramatic changes. Moreover, many existing studies potentially conflate conformity bias with popularity bias and lump together various factors responsible for differences in popularity. In this paper, we identify two distinct types of conformity behavior: informational conformity and normative conformity. To address this, we introduce the TCHN model, which utilizes attentional Hawkes processes to disentangle user self-interest and conformity in a personalized manner. Our approach incorporates temporal graph attention networks to capture users' stable and volatile dynamics. We conduct experiments on three real-world datasets, which uncover diverse levels of conformity among users. The results show that TCHN excels in recommendation accuracy, diversity, and fairness across various user groups.

Abstract Digital libraries aim to provide value to users by housing content that is accessible and searchable. Often such access is afforded through external web search engines. In this article, we measure how easily digital library content can be retrieved (i.e., how retrievable) through a well‐known search engine (Google) using its analytics platforms. Using two measures of document retrievability, we contrast our results with simulation‐based studies that employed synthetic query sets. We determine that estimating the retrievability of content given a Digital Library index is not a strong predictor of how retrievable the content is in practice (via external search engines). Retrievability established the notion that search algorithms can be biased. In our work, we find that while there such bias is present, much of the variation in retrievability appears to be strongly influenced by the queries submitted to the library, a side of retrievability less examined in past work.

With the rapid growth of online misinformation, it is crucial to have reliable fact-checking methods. Recent research on finding check-worthy claims and automated fact-checking have made significant advancements. However, limited guidance exists regarding the presentation of fact-checked content to effectively convey verified information to users. We address this research gap by exploring the critical design elements in fact-checking reports and investigating whether credibility and presentation-based design improvements can enhance users' ability to interpret the report accurately. We co-developed potential content presentation strategies through a workshop involving fact-checking professionals, communication experts, and researchers. The workshop examined the significance and utility of elements such as veracity indicators and explored the feasibility of incorporating interactive components for enhanced information disclosure. Building on the workshop outcomes, we conducted an online experiment involving 76 crowd workers to assess the efficacy of different design strategies. The results indicate that proposed strategies significantly improve users' ability to accurately interpret the verdict of fact-checking articles. Our findings underscore the critical role of effective presentation of fact reports in addressing the spread of misinformation. By adopting appropriate design enhancements, the effectiveness of fact-checking reports can be maximized, enabling users to make informed judgments.

Micro-mobility services (e.g., e-bikes, e-scooters) are increasingly popular among urban communities, being a flexible transport option that brings both opportunities and challenges. As a growing mode of transportation, insights gained from micro-mobility usage data are valuable in policy formulation and improving the quality of services. Existing research analyses patterns and features associated with usage distributions in different localities, and focuses on either temporal or spatial aspects. In this paper, we employ a combination of methods that analyse both spatial and temporal characteristics related to e-scooter trips in a more granular level, enabling observations at different time frames and local geographical zones that prior analysis wasn’t able to do. The insights obtained from anonymised, restricted data on shared e-scooter rides show the applicability of the employed method on regulated, privacy preserving micro-mobility trip data. Our results showed population density is the topmost important feature, and it associates with e-scooter usage positively. Population owning motor vehicles is negatively associated with shared e-scooter trips, suggesting a reduction in e-scooter usage among motor vehicle owners. Furthermore, we found that the effect of humidity is more important than precipitation in predicting hourly e-scooter trip count. Buffer analysis showed, nearly 29% trips were stopped, and 27% trips were started on the footpath, revealing higher utilisation of footpaths for parking e-scooters in Melbourne.

This paper explores the utility of a Large Language Model (LLM) to automatically generate queries and query variants from a description of an information need. Given a set of information needs described as backstories, we explore how similar the queries generated by the LLM are to those generated by humans. We quantify the similarity using different metrics and examine how the use of each set would contribute to document pooling when building test collections. Our results show potential in using LLMs to generate query variants. While they may not fully capture the wide variety of human-generated variants, they generate similar sets of relevant documents, reaching up to 71.1% overlap at a pool depth of 100.

Abstract Knowledge graphs (KGs) are becoming essential resources for many downstream applications. However, their incompleteness may limit their potential. Thus, continuous curation is needed to mitigate this problem. One of the strategies to address this problem is KG alignment, i.e., forming a more complete KG by merging two or more KGs. This paper proposes i-Align, an interpretable KG alignment model. Unlike the existing KG alignment models, i-Align provides an explanation for each alignment prediction while maintaining high alignment performance. Experts can use the explanation to check the correctness of the alignment prediction. Thus, the high quality of a KG can be maintained during the curation process (e.g., the merging process of two KGs). To this end, a novel Transformer-based Graph Encoder (Trans-GE) is proposed as a key component of i-Align for aggregating information from entities’ neighbors (structures). Trans-GE uses Edge-gated Attention that combines the adjacency matrix and the self-attention matrix to learn a gating mechanism to control the information aggregation from the neighboring entities. It also uses historical embeddings, allowing Trans-GE to be trained over mini-batches, or smaller sub-graphs, to address the scalability issue when encoding a large KG. Another component of i-Align is a Transformer encoder for aggregating entities’ attributes. This way, i-Align can generate explanations in the form of a set of the most influential attributes/neighbors based on attention weights. Extensive experiments are conducted to show the power of i-Align. The experiments include several aspects, such as the model’s effectiveness for aligning KGs, the quality of the generated explanations, and its practicality for aligning large KGs. The results show the effectiveness of i-Align in these aspects.

We outline a simulation-based study of the effect rapid population-scale concept drifts have on Collaborative Filtering (CF) models. We create a framework for analyzing the effects of macro-trends in population dynamics on the behavior of such models. Our framework characterizes population-scale concept drifts in item preferences and provides a lens to understand the influence events, such as a pandemic, have on CF models. Our experimental results show the initial impact on CF performance at the initial stage of such events, followed by an aggravated population herding effect during the event. The herding introduces a popularity bias that may benefit affected users, but which comes at the expense of a normal user experience. We propose an adaptive ensemble method that can effectively apply optimal algorithms to cope with the change brought about by different stages of the event.

Where do queries -- the words searchers type into a search box -- come from? The Information Retrieval community understands the performance of queries and search engines extensively, and has recently begun to examine the impact of query variation, showing that different queries for the same information need produce different results. In an information environment where bad actors try to nudge searchers toward misinformation, this is worrisome. The source of query variation -- searcher characteristics, contextual or linguistic prompts, cognitive biases, or even the influence of external parties -- while studied in a piecemeal fashion by other research communities has not been studied by ours. In this paper we draw on a variety of literatures (including information seeking, psychology, and misinformation), and report some small experiments to describe what is known about where queries come from, and demonstrate a clear literature gap around the source of query variations in IR. We chart a way forward for IR to research, document and understand this important question, with a view to creating search engines that provide more consistent, accurate and relevant search results regardless of the searcher's framing of the query.

Non-factoid question answering (NFQA) is a challenging and under-researched task that requires constructing long-form answers, such as explanations or opinions, to open-ended non-factoid questions - NFQs. There is still little understanding of the categories of NFQs that people tend to ask, what form of answers they expect to see in return, and what the key research challenges of each category are.

Concept drift in stream data has been well studied in machine learning applications. In the field of recommender systems, this issue is also widely observed, as known as temporal dynamics in user behavior. Furthermore, in the context of COVID-19 pandemic related contingencies, people shift their behavior patterns extremely and tend to imitate others' opinions. The changes in user behavior may not be always rational. Thus, irrational behavior may impair the knowledge learned by the algorithm. It can cause herd effects and aggravate the popularity bias in recommender systems due to the irrational behavior of users. However, related research usually pays attention to the concept drift of individuals and overlooks the synergistic effect among users in the same social group. We conduct a study on user behavior to detect the collaborative concept drifts among users. Also, we empirically study the increase of experience of individuals can weaken herding effects. Our results suggest the CF models are highly impacted by the herd behavior and our findings could provide useful implications for the design of future recommender algorithms.

Asking clarification questions is an active area of research; however, resources for training and evaluating search clarification methods are not sufficient. To address this issue, we describe MIMICS-Duo, a new freely available dataset of 306 search queries with multiple clarifications (a total of 1,034 query-clarification pairs). MIMICS-Duo contains fine-grained annotations on clarification questions and their candidate answers and enhances the existing MIMICS datasets by enabling multi-dimensional evaluation of search clarification methods, including online and offline evaluation. We conduct extensive analysis to demonstrate the relationship between offline and online search clarification datasets and outline several research directions enabled by MIMICS-Duo. We believe that this resource will help researchers better understand clarification in search.

We examine three statistical significance tests -- a recently proposed ANOVA model and two baseline tests -- using a suite of measures to determine which is better suited for offline evaluation. We apply our analysis to both the runs of a whole TREC track and also to the runs submitted by six participant groups. The former reveals test behavior in the heterogeneous settings of a large-scale offline evaluation initiative; the latter, almost overlooked in past work (to the best of our knowledge), reveals what happens in the much more restricted case of variants of a single system, i.e. the typical context in which companies and research groups operate. We find the ANOVA test strikingly consistent in large-scale settings, but worryingly inconsistent in some participant experiments. Of greater concern, the participant only experiments show one of our baseline tests (a test widely used in research) can produce a substantial number of inconsistent results. We discuss the implications of this inconsistency for possible publication bias.

Abstract This research analyzes human‐generated clarification questions to provide insights into how they are used to disambiguate and provide a better understanding of information needs. A set of clarification questions is extracted from posts on the Stack Exchange platform. Novel taxonomy is defined for the annotation of the questions and their responses. We investigate the clarification questions in terms of whether they add any information to the post (the initial question posted by the asker) and the accepted answer, which is the answer chosen by the asker. After identifying, which clarification questions are more useful, we investigated the characteristics of these questions in terms of their types and patterns. Non‐useful clarification questions are identified, and their patterns are compared with useful clarifications. Our analysis indicates that the most useful clarification questions have similar patterns, regardless of topic. This research contributes to an understanding of clarification in conversations and can provide insight for clarification dialogues in conversational search scenarios and for the possible system generation of clarification requests in information‐seeking conversations.

Taking a user-centric approach, we study the features that render an answer to a non-factoid question useful in the eyes of the person who asked that question. An editorial study, where participants assess the usefulness of the answers they received in response to their questions, as well as 12 different aspects associated with the answers, indicates considerable correlation between certain aspects such as relevance, correctness, and completeness with the user-perceived usefulness of answers. Moreover, we investigate the effectiveness of some commonly used answer quality measures, such as ROGUE, BLEU, METEOR, and BERTScore, demonstrating that these measures are limited in their ability to capture the aspects of usefulness and have room for improvement. The question answering dataset created in our work was made publicly available.

We present a new, multi-faceted taxonomy to classify questions asked in web search engines based on the question intent, types of entities mentioned, types of question words, and granularity of the expected answer. Built based on the inspection of 1,000 real-life questions issued to a web search engine, the taxonomy reflects the recent search behavior of users and enables deep understanding of user intents, goals, and expected answers. This taxonomy is more fine-grained than previous query taxonomies, and is designed with the ultimate goal of reducing the inherent ambiguity in determining the intent of questions. In addition, we describe the formal procedure for conducting an editorial study of the taxonomy including its evaluation. The adopted procedure aims to increase assessor agreement without incurring too much overhead. Our results demonstrate that, despite being more fine-grained, the proposed intent categories result in higher agreement between assessors compared to an existing, commonly used taxonomy.

An introduction to the special issue on conversational search and recommendation is presented in this article. While conversational search and recommendation has roots in early Information Retrieval (IR) research, the recent advances in automatic voice recognition and conversational agents have created increasing interest in this area. In recent years, the IR and related communities have witnessed a number of major contributions to the field of conversational search and recommendation. They include but are not limited to conversational search conceptualization. The growing body of work in this area has been supplemented by an increasing number of recent seminars.

We investigate the impact of popularity bias in false-positive metrics in the offline evaluation of recommender systems. Unlike their true-positive complements, false-positive metrics reward systems that minimize recommendations disliked by users. Our analysis is, to the best of our knowledge, the first to show that false-positive metrics tend to penalise popular items, the opposite behavior of true-positive metrics—causing a disagreement trend between both types of metrics in the presence of popularity biases. We present a theoretical analysis of the metrics that identifies the reason that the metrics disagree and determines rare situations where the metrics might agree—the key to the situation lies in the relationship between popularity and relevance distributions, in terms of their agreement and steepness —two fundamental concepts we formalize. We then examine three well-known datasets using multiple popular true- and false-positive metrics on 16 recommendation algorithms. Specific datasets are chosen to allow us to estimate both biased and unbiased metric values. The results of the empirical study confirm and illustrate our analytical findings. With the conditions of the disagreement of the two types of metrics established, we then determine under which circumstances true-positive or false-positive metrics should be used by researchers of offline evaluation in recommender systems. 1

In this paper we aim to explore the effects of advertisement (ad) quality and position in search engine result pages (SERPs) on a mobile device. We conducted a lab-based eye-tracking study to investigate search time, behavior, and user satisfaction with ads of good or bad qualities positioned at the top or middle of organic results. Our findings suggest that users pay attention to ads regardless of their quality or position. However, they tend to pay different amounts of attention to organic results and SERPs because of ad quality. We also found that user satisfaction and the chance of clicking on an ad vary according to ad quality and position.

To study the impact of providing direct answers in search results on user behavior, we conducted a controlled user study to analyze factors including reading time, eye-tracked attention, and the influence of the quality of answer module content. We also studied a more advanced answer interface, where multiple answers are shown on the search engine results page (SERP). Our results show that users focus more extensively than normal on the top items in the result list when answers are provided. The existence of the answer module helps to improve user engagement on SERPs, reduces user effort, and promotes user satisfaction during the search process. Furthermore, we investigate how the question type -- factoid or non-factoid -- affects user interaction patterns. This work provides insight into the design of SERPs that includes direct answers to queries, including when answers should be shown.

We investigated how users evaluate passage-length answers for non-factoid questions. We conduct a study where answers were presented to users, sometimes shown with automatic word highlighting. Users were tasked with evaluating answer quality, correctness, completeness, and conciseness. Words in the answer were also annotated, both explicitly through user mark up and implicitly through user gaze data obtained from eye-tracking. Our results show that the correctness of an answer strongly depends on its completeness, conciseness is less important.

We introduce the novel research problem of task recognition in daily life. We recognize tasks such as project management, planning, meal-breaks, communication, documentation, and family care. We capture Cyber, Physical, and Social (CPS) activities of 17 participants over four weeks using device-based sensing, app activity logging, and an experience sampling methodology. Our cohort includes students, casual workers, and professionals, forming the first real-world context-rich task behaviour dataset. We model CPS activities across different task categories, results highlight the importance of considering the CPS feature sets in modelling, especially work-related tasks.

False-positive metrics can capture an important side of recommendation quality, focusing on the impact of suggestions that are disliked by users, as a complement of common metrics that only measure the amount of successful recommendations. In this paper we research the extent to which false-positive metrics agree or disagree with true-positive metrics in the offline evaluation of recommender systems. We discover a surprising degree of systematic disagreement that was occasionally noted but not explained in the literature by previous authors. We find an explanation for the discrepancy be-tween the metrics in the effect of popularity biases, which impact false and true-positive metrics in very different ways: instead of rewarding the recommendation of popular items, as with true-positive, false-positive metrics penalize the popular. We determine precise conditions and cases in the general trends, with a formal explanation for our findings, which we confirm and illustrate empirically in experiments with different datasets.

Extractive-abstractive hybrid summarization can generate readable, concise summaries for long documents. Extraction-then-abstraction and extraction-with-abstraction are two representative approaches to hybrid summarization. But their general performance is yet to be evaluated by large scale experiments.We examined two state-of-the-art hybrid summarization algorithms from three novel perspectives: we applied them to a form of headline generation not previously tried, we evaluated the generalization of the algorithms by testing them both within and across news domains; and we compared the automatic assessment of the algorithms to human comparative judgments. It is found that an extraction-then-abstraction hybrid approach outperforms an extraction-with-abstraction approach, particularly for cross-domain headline generation.

Examination is one of the most important user interactions in Web search. A number of works studied examination behavior in Web search and helped researchers better understand how users allocate their attention on search engine result pages (SERPs). Compared to desktop search, mobile search has a number of differences such as fewer results on the screen. These differences bring in mobile-specific factors affecting users' examination behavior. However, there still lacks research on users' attention allocation mechanism via viewports in mobile search. Therefore, we design a lab-based study to collect user's rich interaction behavior in mobile search. Based on the collected data, we first analyze how users examine SERPs and allocate their attention to heterogeneous results. Then we investigate the effect of mobile-specific factors and other common factors on users allocating attention. Finally, we apply the findings of user attention allocation from the user study into click model construction efforts, which significantly improves the state-of-the-art click model. Our work brings insights into a better understanding of users' interaction patterns in mobile search and may benefit other mobile search-related research.

Intelligent assistants can serve many purposes, including entertainment (e.g. playing music), home automation, and task management (e.g. timers, reminders). The role of these assistants is evolving to also support people engaged in work tasks, in workplaces and beyond. To design truly useful intelligent assistants for work, it is important to better understand the work tasks that people are performing. Based on a survey of 401 respondents' daily tasks and activities in a work setting, we present a classification of work-related tasks, and analyze their key characteristics, including the frequency of their self-reported tasks, the environment in which they undertake the tasks, and which, if any, electronic devices are used. We also investigate the cyber, physical, and social aspects of tasks. Finally, we reflect on how intelligent assistants could influence and help people in a work environment to complete their tasks, and synthesize our findings to provide insight on the future of intelligent assistants in support of amplifying personal productivity.

We investigated the learning process in search by conducting a log-based study involving registered job seekers of a commercial job search engine. The analysis shows that job search is a complex task: seekers usually submit multiple queries over sessions that can last days or even weeks. We find that querying, clicking, and job application rates change over time: job seekers tend to use more filters and a less diverse set of query terms. In terms of click and application behavior, we observed a significant decrease in click rate and query term diversity, as well as an increase in application rates. These trends are found to largely match information seeking models of learning in a complex search task. However, common behaviors are observed in the logs that suggest the existing models may not be sufficient to describe all of the users' learning and seeking processes.

We learn user preferences from ratings and reviews by using multi-task learning (MTL) of rating prediction and summarization of item reviews. Reviews of an item tend to describe detailed user preferences (e.g., the cast, genre, or screenplay of a movie). A summary of such a review or a rating describes an overall user experience of the item. Our objective is to learn latent vectors which are shared across rating prediction and review summary generation. Additionally, the learned latent vectors and the generated summary act as explanations for the recommendation. Our MTL-based approach J3R uses a multi-layer perceptron for rating prediction, combined with pointer-generator networks with attention mechanism for the summarization component. We provide empirical evidence for joint learning of rating prediction and summary generation being beneficial for recommendation by conducting experiments on the Yelp dataset and six domains of the Amazon 5-core dataset. Additionally, we provide two ways of explanations visualizing (a) the user vectors on different topics of a domain, computed from our J3R approach and (b) a ten-word review summary of a review and the attention highlights generated on the review based on the user-item vectors

We improve the measurement accuracy of retrieval system performance by better modeling the noise present in test collection scores. Our technique draws its inspiration from two approaches: one, which exploits the variable measurement accuracy of topics; the other, which randomly splits document collections into shards. We describe and theoretically analyze an ANOVA model able to capture the effects of topics, systems, and document shards as well as their interactions. Using multiple TREC collections, we empirically confirm theoretical results in terms of improved estimation accuracy and robustness of found significant differences. The improvements compared to widely used test collection measurement techniques are substantial. We speculate that our technique works because we do not assume that the topics of a test collection measure performance equally.

The task intelligence workshop at the 2019 ACM Web Search and Data Mining (WSDM) conference comprised a mixture of research paper presentations, reports from data challenge participants, invited keynote(s) on broad topics related to tasks, and a workshop-wide discussion about task intelligence and its implications for system development.

Despite the bulk of research studying how to more accurately compare the performance of IR systems, less attention is devoted to better understanding the different factors that play a role in such performance and how they interact. This is the case of shards, i.e., partitioning a document collection into sub-parts, which are used for many different purposes, ranging from efficiency to selective search or making test collection evaluation more accurate. In all these cases, there is empirical knowledge supporting the importance of shards, but we lack actual models that allow us to measure the impact of shards on system performance and how they interact with topics and systems. We use the general linear mixed model framework and present a model that encompasses the experimental factors of system, topic, shard, and their interaction effects. This detailed model allows us to more accurately estimate differences between the effect of various factors. We study shards created by a range of methods used in prior work and better explain observations noted in prior work in a principled setting and offer new insights. Notably, we discover that the topic*shard interaction effect, in particular, is a large effect almost globally across all datasets, an observation that, to our knowledge, has not been measured before.

We analyze fifteen Twitter user geolocation models and two baselines comparing how they are evaluated. Our results demonstrate that the choice of effectiveness metric can have a substantial impact on the conclusions drawn from an experiment. We show that for general evaluations, a range of metrics should be reported to ensure that a complete picture of system effectiveness is conveyed.

Typing is a common form of query input for search engines and other information retrieval systems; we therefore investigate the relationship between typing behavior and search interactions. The search process is interactive and typically requires entering one or more queries, and assessing both summaries from Search Engine Result Pages and the underlying documents, to ultimately satisfy some information need. Under the Search Economic Theory model of interactive information retrieval, differences in query costs will result in search behavior changes. We investigate how differences in query inputs themselves may relate to Search Economic Theory by conducting a lab-based experiment to observe how text entries influence subsequent search interactions. Our results indicate that for faster typing speeds, more queries are entered in a session, while both query lengths and assessment times are lower.

We conducted a laboratory-based observational study where pairs of people performed search tasks communicating verbally. Examination of the discourse allowed commonly used interactions to be identified for Spoken Conversational Search (SCS). We compared the interactions to existing models of search behaviour. We find that SCS is more complex and interactive than traditional search. This work enhances our understanding of different search behaviours and proposes research opportunities for an audio-only search system. Future work will focus on creating models of search behaviour for SCS and evaluating these against actual SCS systems.

Errors in formulation of queries made by users can lead to poor search results pages. We performed a living lab study using online A/B testing to measure the degree of improvement achieved with a query amendment technique when applied to a commercial job search engine. Of particular interest in this case study is a clear 'success' signal, namely, the number of job applications lodged by a user as a result of querying the service. A set of 276 queries was identified for amendment in four different categories through the use of word embeddings, with large gains in conversion rates being attained in all four of those categories. Our analysis of query reformulations also provides a better understanding of user satisfaction in the case of problematic queries (ones with fewer results than fill a single page) by observing that users tend to reformulate rewritten queries less.

Evidence derived from passages that closely represent likely answers to a posed query can be useful input to the ranking process. Based on a novel use of Community Question Answering data, we present an approach for the creation of such passages. A general framework for extracting answer passages and estimating their quality is proposed, and this evidence is integrated into ranking models. Our experiments on two web collections show that such quality estimates from answer passages provide a strong indication of document relevance and compare favorably to previous passage-based methods. Combining such evidence can significantly improve over a set of state-of-the-art ranking models, including Quality-Biased Ranking, External Expansion, and a combination of both. A final ranking model that incorporates all quality estimates achieves further improvements on both collections.

Traditionally, recommender systems modelled the physical and cyber contextual influence on people's moving, querying, and browsing behaviors in isolation. Yet, searching, querying, and moving behaviors are intricately linked, especially indoors. Here, we introduce a tripartite location-query-browse graph (LQB) for nuanced contextual recommendations. The LQB graph consists of three kinds of nodes: locations, queries, and Web domains. Directed connections only between heterogeneous nodes represent the contextual influences, while connections of homogeneous nodes are inferred from the contextual influences of the other nodes. This tripartite LQB graph is more reliable than any monopartite or bipartite graph in contextual location, query, and Web content recommendations. We validate this LQB graph in an indoor retail scenario with extensive dataset of three logs collected from over 120,000 anonymized, opt-in users over a 1-year period in a large inner-city mall in Sydney, Australia. We characterize the contextual influences that correspond to the arcs in the LQB graph, and evaluate the usefulness of the LQB graph for location, query, and Web content recommendations. The experimental results show that the LQB graph successfully captures the contextual influence and significantly outperforms the state of the art in these applications.

We formulate a document summarization method to extract passage-level answers for non-factoid queries, referred to as answer-biased summaries. We propose to use external information from related Community Question Answering (CQA) content to better identify answer bearing sentences. Three optimization-based methods are proposed: (i) query-biased, (ii) CQA-answer-biased, and (iii) expanded-query-biased, where expansion terms were derived from related CQA content. A learning-to-rank-based method is also proposed that incorporates a feature extracted from related CQA content. Our results show that even if a CQA answer does not contain a perfect answer to a query, their content can be exploited to improve the extraction of answer-biased summaries from other corpora. The quality of CQA content is found to impact on the accuracy of optimization-based summaries, though medium quality answers enable the system to achieve a comparable (and in some cases superior) accuracy to state-of-the-art techniques. The learning-to-rank-based summaries, on the other hand, are not significantly influenced by CQA quality. We provide a recommendation of the best use of our proposed approaches in regard to the availability of different quality levels of related CQA content. As a further investigation, the reliability of our approaches was tested on another publicly available dataset.

We present preliminary findings from a study of mixed initiative conversational behaviour for informational search in an acoustic setting. The aim of the observational study is to reveal insights into how users would conduct searches over voice where a screen is absent but where users are able to converse interactively with the search system. We conducted alaboratory-based observational study of 13 pairs of participants each completing three search tasks with different cognitive complexity levels. The communication between the pairs was analyzed for interaction patterns used in the search process. This setup mimics the situation of a user interacting with a search system via a speech-only interface.

We study a new type of spatial-textual trajectory search: the Exemplar Trajectory Query (ETQ), which specifies one or more places to visit, and descriptions of activities at each place. Our goal is to efficiently find the top-k trajectories by computing spatial and textual similarity at each point. The computational cost for pointwise matching is significantly higher than previous approaches. Therefore, we introduce an incremental pruning baseline and explore how to adaptively tune our approach, introducing a gap-based optimization and a novel twolevel threshold algorithm to improve efficiency. Our proposed methods support order-sensitive ETQ with a minor extension. Experiments on two datasets verify the efficiency and scalability of our proposed solution.

Incorporating conventional, unsupervised features into a neural architecture has the potential to improve modeling effectiveness, but this aspect is often overlooked in the research of deep learning models for information retrieval. We investigate this incorporation in the context of answer sentence selection, and show that combining a set of query matching, readability, and query focus features into a simple convolutional neural network can lead to markedly increased effectiveness. Our results on two standard question-answering datasets show the effectiveness of the combined model.

Understanding the factors comprising IR system effectiveness is of primary importance to compare different IR systems. Effectiveness is traditionally broken down, using ANOVA, into a topic and a system effect but this leaves out a key component of our evaluation paradigm: the collections of documents. We break down effectiveness into topic, system and sub-corpus effects and compare it to the traditional break down, considering what happens when different evaluation measures come into play. We found that sub-corpora are a significant effect. The consideration of which allows us to be more accurate in estimating what systems are significantly different. We also found that the sub-corpora affect different evaluation measures in different ways and this may impact on what systems are considered significantly different.

This paper investigates if Information Foraging Theory can be used to understand differences in user behavior when searching on mobile and desktop web search systems. Two groups of thirty-six participants were recruited to carry out six identical web search tasks on desktop or on mobile. The search tasks were prepared with a different number and distribution of relevant documents on the first result page. Search behaviors on mobile and desktop were measurably different. Desktop participants viewed and clicked on more results but saved fewer as relevant, compared to mobile participants, when information scent level increased. Mobile participants achieved higher search accuracy than desktop participants for tasks with increasing numbers of relevant search results. Conversely, desktop participants were more accurate than mobile participants for tasks with an equal number of relevant results that were more distributed across the results page. Overall, both an increased number and better positioning of relevant search results improved the ability of participants to locate relevant results on both desktop and mobile. Participants spent more time and issued more queries on desktop, but abandoned less and saved more results for initial queries on mobile.

The increase of voice-based interaction has changed the way people seek information, making search more conversational. Development of effective conversational approaches to search requires better understanding of how people express information needs in dialogue. This paper describes the creation and examination of over 32K spoken utterances collected during 34 hours of collaborative search tasks. The contribution of this work is three-fold. First, we propose a model of conversational information needs (CINs) based on a synthesis of relevant theories in Information Seeking and Retrieval. Second, we show several behavioural patterns of CINs based on the proposed model. Third, we identify effective feature groups that may be useful for detecting CINs categories from conversations. This paper concludes with a discussion of how these findings can facilitate advance of conversational search applications.

We analyze 18‐ million rows of Wi‐Fi access logs collected over a 1‐year period from over 120,000 anonymized users at an inner city shopping mall. The anonymized data set gathered from an opt‐in system provides users' approximate physical location as well as web browsing and some search history. Such data provide a unique opportunity to analyze the interaction between people's behavior in physical retail spaces and their web behavior, serving as a proxy to their information needs. We found that (a) there is a weekly periodicity in users' visits to the mall; (b) people tend to visit similar mall locations and web content during their repeated visits to the mall; (c) around 60% of registered Wi‐Fi users actively browse the web, and around 10% of them use Wi‐Fi for accessing web search engines; (d) people are likely to spend a relatively constant amount of time browsing the web while the duration of their visit may vary; (e) the physical spatial context has a small, but significant, influence on the web content that indoor users browse; and (f) accompanying users tend to access resources from the same web domains.

We address the challenge of extracting query biased audio summaries from podcasts to support users in making relevance decisions in spoken document search via an audio‐only communication channel. We performed a crowdsourced experiment that demonstrates that transcripts of spoken documents created using Automated Speech Recognition (ASR), even with significant errors, are effective sources of document summaries or “snippets” for supporting users in making relevance judgments against a query. In particular, the results show that summaries generated from ASR transcripts are comparable, in utility and user‐judged preference, to spoken summaries generated from error‐free manual transcripts of the same collection. We also observed that content‐based audio summaries are at least as preferred as synthesized summaries obtained from manually curated metadata, such as title and description. We describe a methodology for constructing a new test collection, which we have made publicly available.

Re-finding is the process of searching for information that a user has previously encountered and is a common activity carried out with information retrieval systems. In this work, we investigate re-finding in the context of vertical search, differentiating and modeling user re-finding behavior within different media and topic domains, including images, news, reference material, and movies. We distinguish the re-finding behavior in vertical domains from re-finding in a general search context and engineer features that are effective in differentiating re-finding across the domains. The features are then used to build machine-learned models, achieving an accuracy of re-finding detection in verticals of 85.7% on average. Our results demonstrate that detecting re-finding in specific verticals is more difficult than examining re-finding for general search tasks. We then investigate the effectiveness of differentiating re-finding behavior in two restricted contexts: We consider the case where the history of a searcher’s interactions with the search system is not available. In this scenario, our features and models achieve an average accuracy of 77.5% across the domains. We then examine the detection of re-finding during the early part of a search session. Both of these restrictions represent potential real-world search scenarios, where a system is attempting to learn about a user but may have limited information available. Finally, we investigate in which types of domains re-finding is most difficult. Here, it would appear that re-finding images is particularly challenging for users. This research has implications for search engine design, in terms of adapting search results by predicting the type of user tasks and potentially enabling the presentation of vertical-specific results when re-finding is identified. To the best of our knowledge, this is the first work to investigate the issue of vertical re-finding.

Query-level instance weighting is a technique for unsupervised transfer ranking, which aims to train a ranker on a source collection so that it also performs effectively on a target collection, even if no judgement information exists for the latter. Past work has shown that this approach can be used to significantly improve effectiveness; in this work, the approach is re-examined on a wide set of publicly available L2R test collections with more advanced learning to rank algorithms. Different query-level weighting strategies are examined against two transfer ranking frameworks: AdaRank and a new weighted LambdaMART algorithm. Our experimental results show that the effectiveness of different weighting strategies, including those shown in past work, vary under different transferring environments. In particular, (i) Kullback-Leibler based density-ratio estimation tends to outperform a classification-based approach and (ii) aggregating document-level weights into query-level weights is likely superior to direct estimation using a query-level representation. The Nemenyi statistical test, applied across multiple datasets, indicates that most weighting transfer learning methods do not significantly outperform baselines, although there is potential for the further development of such techniques.

Information Retrieval (IR) research has traditionally focused on serving the best results for a single query - so-called ad hoc retrieval. However, users typically search iteratively, refining and reformulating their queries during a session. A key challenge in the study of this interaction is the creation of suitable evaluation resources to assess the effectiveness of IR systems over sessions. This paper describes the TREC Session Track, which ran from 2010 through to 2014, which focussed on forming test collections that included various forms of implicit feedback. We describe the test collections; a brief analysis of the differences between datasets over the years; and the evaluation results that demonstrate that the use of user session data significantly improved effectiveness.

The International Classification of Diseases (ICD) is a type of meta-data found in many Electronic Patient Records. Research to explore the utility of these codes in medical Information Retrieval (IR) applications is new, and many areas of investigation remain, including the question of how reliable the assignment of the codes has been. This paper proposes two uses of the ICD codes in two different contexts of search: Pseudo-Relevance Judgments (PRJ) and Pseudo-Relevance Feedback (PRF). We find that our approach to evaluate the TREC challenge runs using simulated relevance judgments has a positive correlation with the TREC official results, and our proposed technique for performing PRF based on the ICD codes significantly outperforms a traditional PRF approach. The results are found to be consistent over the two years of queries from the TREC medical test collection.

Finding a place of interest (e.g., a restaurant, hotel, or attraction) is often related to a group information need, however, the actual multiparty collaboration in such searches has not been explored, and little is known about its significance and related practices. We surveyed 100 computer science students and found that 94% (of respondents) searched for places online; 87% had done so as part of a group. Search for place by multiple active participants was experienced by 78%, with group sizes typically being 2 or 3. Search occurred in a range of settings with both desktop PCs and mobile devices. Difficulties were reported with coordinating tasks, sharing results, and making decisions. The results show that finding a place of interest is a quite different group‐based search than other multiparty information‐seeking activities. The results suggest that local search systems, their interfaces and the devices that access them can be made more usable for collaborative search if they include support for coordination, sharing of results, and decision making.

We examined whether the microblog comments given by people after reading a web document could be exploited to improve the accuracy of a web document summarization system. We examined the effect of social information (i.e., tweets) on the accuracy of the generated summaries by comparing the user preference for TBS (tweet‐biased summary) with GS (generic summary). The result of crowdsourcing‐based evaluation shows that the user preference for TBS was significantly higher than GS. We also took random samples of the documents to see the performance of summaries in a traditional evaluation using ROUGE, which, in general, TBS was also shown to be better than GS. We further analyzed the influence of the number of tweets pointed to a web document on summarization accuracy, finding a positive moderate correlation between the number of tweets pointed to a web document and the performance of generated TBS as measured by user preference. The results show that incorporating social information into the summary generation process can improve the accuracy of summary. The reason for people choosing one summary over another in a crowdsourcing‐based evaluation is also presented in this article.

We present a study of which baseline to use when testing a new retrieval technique. In contrast to past work, we show that measuring a statistically significant improvement over a weak baseline is not a good predictor of whether a similar improvement will be measured on a strong baseline. Sometimes strong baselines are made worse when a new technique is applied. We investigate whether conducting comparisons against a range of weaker baselines can increase confidence that an observed effect will also show improvements on a stronger baseline. Our results indicate that this is not the case -- at best, testing against a range of baselines means that an experimenter can be more confident that the new technique is unlikely to significantly harm a strong baseline. Examining recent past work, we present evidence that the information retrieval (IR) community continues to test against weak baselines. This is unfortunate as, in light of our experiments, we conclude that the only way to be confident that a new technique is a contribution is to compare it against nothing less than the state of the art.

Many IR effectiveness measures are motivated from intuition, theory, or user studies. In general, most effectiveness measures are well correlated with each other. But, what about where they don't correlate? Which rankings cause measures to disagree? Are these rankings predictable for particular pairs of measures? In this work, we examine how and where metrics disagree, and identify differences that should be considered when selecting metrics for use in evaluating retrieval systems.

Presenting search results over a speech-only communication channel involves a number of challenges for users due to cognitive limitations and the serial nature of speech. We investigated the impact of search result summary length in speech-based web search, and compared our results to a text baseline. Based on crowdsourced workers, we found that users preferred longer, more informative summaries for text presentation. For audio, user preferences depended on the style of query. For single-facet queries, shortened audio summaries were preferred, additionally users were found to judge relevance with a similar accuracy compared to text-based summaries. For multi-facet queries, user preferences were not as clear, suggesting that more sophisticated techniques are required to handle such queries.

The question of which type of computer science (CS) publication—conference or journal—is likely to result in more citations for a published paper is addressed. A series of data sets are examined and joined in order to analyze the citations of over 195,000 conference papers and 108,000 journal papers. Two means of evaluating the citations of journals and conferences are explored: h5 and average citations per paper; it was found that h5 has certain biases that make it a difficult measure to use (despite it being the main measure used by Google Scholar). Results from the analysis show that CS, as a discipline, values conferences as a publication venue more highly than any other academic field of study. The analysis also shows that a small number of elite CS conferences have the highest average paper citation rate of any publication type, although overall, citation rates in conferences are no higher than in journals. It is also shown that the length of a paper is correlated with citation rate.

Past work showed that significant inconsistencies between retrieval results occurred on different test collections, even when one of the test collections contained only a subset of the documents in the other. However, the experimental methodologies in that paper made it hard to determine the cause of the inconsistencies. Using a novel methodology that eliminates the problems with uneven distribution of relevant documents, we confirm that observing a statistically significant improvement between two IR systems can be strongly influenced by the choice of documents in the test collection. We investigate two possible causes of this problem of test collections. Our results show that collection size and document source have a strong influence in the way that a test collection will rank one retrieval system relative to another. This is of particular interest when constructing test collections, as we show that using different subsets of a collection produces differing evaluation results.

We report a preliminary study of mobile Web behaviour in a large indoor retail space. By analysing a Web log collected over a 1 year period at an inner city shopping mall in Sydney, Australia, we found that 1) around 60% of registered Wi-Fi users actively browse the Internet, and the rest 40% do not, with around 10% of these users using Web search engines. Around 70% of this Web activity in the investigated mall come from frequent visitors; 2) the content that indoor users search for is different from the content they consume while browsing; 3) the popularity of future indoor search queries can be predicted with a simple theoretical model based on past queries treated as a weighted directed graph. The work described in this paper underpins applications such as the prediction of users' information needs, retail recommendation systems, and improving the mobile Web search experience.

In this work, we investigate approaches to engineer better topic sets in information retrieval test collections. By recasting the TREC evaluation exercise from one of building more effective systems to an exercise in building better topics, we present two possible approaches to quantify topic "goodness": topic ease and topic set predictivity. A novel interpretation of a well known result and a twofold analysis of data from several TREC editions lead to a result that has been neglected so far: both topic ease and topic set predictivity have changed significantly across the years, sometimes in a perhaps undesirable way.

Information retrieval test collections traditionally use a combination of automatic and manual runs to create a pool of documents to be judged. The quality of the final judgments produced for a collection is a product of the variety across each of the runs submitted and the pool depth. In this work, we explore fully automated approaches to generating a pool. By combining a simple voting approach with machine learning from documents retrieved by automatic runs, we are able to identify a large portion of relevant documents that would normally only be found through manual runs. Our initial results are promising and can be extended in future studies to help test collection curators ensure proper judgment coverage is maintained across complete document collections.

The use of sampling, randomized algorithms, or training based on the unpredictable inputs of users in Information Retrieval often leads to non-deterministic outputs. Evaluating the effectiveness of systems incorporating these methods can be challenging since each run may produce different effectiveness scores. Current IR evaluation techniques do not address this problem. Using the context of distributed information retrieval as a case study for our investigation, we propose a solution based on multivariate linear modeling. We show that the approach provides a consistent and reliable method to compare the effectiveness of non-deterministic IR algorithms, and explain how statistics can safely be used to show that two IR algorithms have equivalent effectiveness.

We investigate the application of a light-weight approach to result list clustering for the purposes of diversifying search results. We introduce a novel post-retrieval approach, which is independent of external information or even the full-text content of retrieved documents; only the retrieval score of a document is used. Our experiments show that this novel approach is beneficial to effectiveness, albeit only on certain baseline systems. The fact that the method works indicates that the retrieval score is potentially exploitable in diversity.

Search engine users typically engage in multiquery sessions in their quest to fulfill their information needs. Despite a plethora of research findings suggesting that a significant group of users look for information within a specific geographical scope, existing reformulation studies lack a focused analysis of how users reformulate geographic queries. This study comprehensively investigates the ways in which users reformulate such needs in an attempt to fill this gap in the literature. Reformulated sessions were sampled from a query log of a major search engine to extract 2,400 entries that were manually inspected to filter geo sessions. This filter identified 471 search sessions that included geographical intent, and these sessions were analyzed quantitatively and qualitatively. The results revealed that one in five of the users who reformulated their queries were looking for geographically related information. They reformulated their queries by changing the content of the query rather than the structure. Users were not following a unified sequence of modifications and instead performed a single reformulation action. However, in some cases it was possible to anticipate their next move. A number of tasks in geo modifications were identified, including standard, multi‐needs, multi‐places, and hybrid approaches. The research concludes that it is important to specialize query reformulation studies to focus on particular query types rather than generically analyzing them, as it is apparent that geographic queries have their special reformulation characteristics.

In this paper we present results from an investigation of religious information searching based on analyzing log files from a large general‐purpose search engine. From approximately 15 million queries, we identified 124,422 that were part of 60,759 user sessions. We present a method for categorizing queries based on related terms and show differences in search patterns between religious searches and web searching more generally. We also investigate the search patterns found in queries related to 5 religions: Christianity,Hinduism,Islam,Buddhism, andJudaism. Different search patterns are found to emerge. Results from this study complement existing studies of religious information searching and provide a level of detailed analysis not reported to date. We show, for example, that sessions involving religion‐related queries tend to last longer, that the lengths of religion‐related queries are greater, and that the number of uniqueURLsclicked is higher when compared to all queries. The results of the study can serve to provide information on what this large population of users is actually searching for.

The relative performance of retrieval systems when evaluated on one part of a test collection may bear little or no similarity to the relative performance measured on a different part of the collection. In this paper we report the results of a detailed study of the impact that different sub-collections have on retrieval effectiveness, analyzing the effect over many collections, and with different approaches to sub-dividing the collections. The effect is shown to be substantial, impacting on comparisons between retrieval runs that are statistically significant. Some possible causes for the effect are investigated, and the implications of this work are examined for test collection design and for the strength of conclusions one can draw from experimental results.

The standard system-based evaluation paradigm has focused on assessing the performance of retrieval systems in serving the best results for a single query. Real users, however, often begin an interaction with a search engine with a sufficiently under-specified query that they will need to reformulate before they find what they are looking for. In this work we consider the problem of evaluating retrieval systems over test collections of multi-query sessions. We propose two families of measures: a model-free family that makes no assumption about the user's behavior over a session, and a model-based family with a simple model of user interactions over the session. In both cases we generalize traditional evaluation metrics such as average precision to multi-query session evaluation. We demonstrate the behavior of the proposed metrics by using the new TREC 2010 Session track collection and simulations over the TREC-9 Query track collection.

Relevance assessments are a key component for test collection-based evaluation of information retrieval systems. This paper reports on a feature of such collections that is used as a form of ground truth data to allow analysis of human assessment error. A wide range of test collections are retrospectively examined to determine how accurately assessors judge the relevance of documents. Our results demonstrate a high level of inconsistency across the collections studied. The level of irregularity is shown to vary across topics, with some showing a very high level of assessment error. We investigate possible influences on the error, and demonstrate that inconsistency in judging increases with time. While the level of detail in a topic specification does not appear to influence the errors that assessors make, judgements are significantly affected by the decisions made on previously seen similar documents. Assessors also display an assessment inertia. Alternate approaches to generating relevance judgements appear to reduce errors. A further investigation of the way that retrieval systems are ranked using sets of relevance judgements produced early and late in the judgement process reveals a consistent influence measured across the majority of examined test collections.

The emergence of large-scale social Web communities has enabled users to share online vast amounts of multimedia content. An analysis of YouTube reveals a high amount of redundancy, in the form of videos with overlapping or duplicated content. We use robust content-based video analysis techniques to detect overlapping sequences between videos. Based on the output of these techniques, we present an in-depth study of duplication and content overlap in YouTube, and analyze various dependencies between content overlap and meta data such as video titles, views, video ratings, and tags. As an application, we show that content-based links provide useful information for generating new tag assignments. We propose different tag propagation methods for automatically obtaining richer video annotations. Experiments on video clustering and classification as well as a user evaluation demonstrate the viability of our approach.

This paper presents results comparing user preference for search engine rankings with measures of effectiveness computed from a test collection. It establishes that preferences and evaluation measures correlate: systems measured as better on a test collection are preferred by users. This correlation is established for both "conventional web retrieval" and for retrieval that emphasizes diverse results. The nDCG measure is found to correlate best with user preferences compared to a selection of other well known measures. Unlike previous studies in this area, this examination involved a large population of users, gathered through crowd sourcing, exposed to a wide range of retrieval systems, test collections and search tasks. Reasons for user preferences were also gathered and analyzed. The work revealed a number of new results, but also showed that there is much scope for future work refining effectiveness measures to better capture user preferences.

Abstract The authors investigate factors influencing user satisfaction in information retrieval. It is evident from this study that user satisfaction is a subjective variable, which can be influenced by several factors such as system effectiveness, user effectiveness, user effort, and user characteristics and expectations. Therefore, information retrieval evaluators should consider all these factors in obtaining user satisfaction and in using it as a criterion of system effectiveness. Previous studies have conflicting conclusions on the relationship between user satisfaction and system effectiveness; this study has substantiated these findings and supports using user satisfaction as a criterion of system effectiveness.

Abstract Folder navigation is the main way that personal computer users retrieve their own files. People dedicate considerable time to creating systematic structures to facilitate such retrieval. Despite the prevalence of both manual organization and navigation, there is very little systematic data about how people actually carry out navigation, or about the relation between organization structure and retrieval parameters. The aims of our research were therefore to study users' folder structure, personal file navigation, and the relations between them. We asked 296 participants to retrieve 1,131 of their active files and analyzed each of the 5,035 navigation steps in these retrievals. Folder structures were found to be shallow (files were retrieved from mean depth of 2.86 folders), with small folders (a mean of 11.82 files per folder) containing many subfolders (M=10.64). Navigation was largely successful and efficient with participants successfully accessing 94% of their files and taking 14.76 seconds to do this on average. Retrieval time and success depended on folder size and depth. We therefore found the users' decision to avoid both deep structure and large folders to be adaptive. Finally, we used a predictive model to formulate the effect of folder depth and folder size on retrieval time, and suggested an optimization point in this trade‐off.

Abstract This study investigated a relatively unexamined query type, queries composed of URLs. The extent, variation, and user click‐through behavior was examined to determine the intent behind URL queries. The study made use of a search log from which URL queries were identified and selected for both qualitative and quantitative analyses. It was found that URL queries accounted for ∼17% of the sample. There were statistically significant differences between URL queries and non‐URL queries in the following attributes: mean query length; mean number of tokens per query; and mean number of clicks per query. Users issuing such queries clicked on fewer result list items higher up the ranking compared to non‐URL queries. Classification indicated that nearly 86% of queries were navigational in intent with informational and transactional queries representing about 7% of URL queries each. This is in contrast to past research that suggested that URL queries were 100% navigational. The conclusions of this study are that URL queries are relatively common and that simply returning the page that matches a user's URL is not an optimal strategy.

In this paper we examine user queries with respect to diversity: providing a mix of results across different interpretations. Using two query log analysis techniques (click entropy and reformulated queries), 14.9 million queries from the Microsoft Live Search log were analysed. We found that a broad range of query types may benefit from diversification. Additionally, although there is a correlation between word ambiguity and the need for diversity, the range of results users may wish to see for an ambiguous query stretches well beyond traditional notions of word sense.

The analysis of the leading social video sharing platform YouTube reveals a high amount of redundancy, in the form of videos with overlapping or duplicated content. In this paper, we show that this redundancy can provide useful information about connections between videos. We reveal these links using robust content-based video analysis techniques and exploit them for generating new tag assignments. To this end, we propose different tag propagation methods for automatically obtaining richer video annotations. Our techniques provide the user with additional information about videos, and lead to enhanced feature representations for applications such as automatic data organization and search. Experiments on video clustering and classification as well as a user evaluation demonstrate the viability of our approach.

Abstract This brief communication describes the results of a questionnaire examining certain aspects of the Web‐based information seeking practices of university students. The results are contrasted with past work showing that queries to Web search engines can be assigned to one of a series of categories: navigational, informational, and transactional. The survey results suggest that a large group of queries, which in the past would have been classified as informational, have become at least partially navigational. We contend that this change has occurred because of the rise of large Web sites holding particular types of information, such as Wikipedia and the Internet Movie Database.

This paper investigates the agreement of relevance assessments between official TREC judgments and those generated from an interactive IR experiment. Results show that 63% of documents judged relevant by our users matched official TREC judgments. Several factors contributed to differences in the agreements: the number of retrieved relevant documents; the number of relevant documents judged; system effectiveness per topic and the ranking of relevant documents.

Test collections are extensively used in the evaluation of information retrieval systems. Crucial to their use is the degree to which results from them predict user effectiveness. At first, past studies did not substantiate a relationship between system and user effectiveness; more recently, however, correlations have begun to emerge. The results of this paper strengthen and extend those findings. We introduce a novel methodology for investigating the relationship, which shows great success in establishing a significant correlation between system and user effectiveness. It is shown that users behave differently and discern differences between pairs of systems that have a very small absolute difference in test collection effectiveness. Our results strengthen the use of test collections in IR evaluation, confirming that users' effectiveness can be predicted successfully.

Although there are many papers examining ambiguity in Information Retrieval, this paper shows that there is a whole class of ambiguous word that past research has barely explored. It is shown that the class is more ambiguous than other word types and is commonly used in queries. The lack of test collections containing ambiguous queries is highlighted and a method for creating collections from existing resources is described. Tests using the new collection show the impact of query ambiguity on an IR system: it is shown that conventional systems are incapable of dealing effectively with such queries and that current assumptions about how to improve search effectiveness do not hold when searching on this common query type.

Abstract A brief communication appearing in this journal ranked UK‐based LIS and (some) IR academics by theirh‐indexusing data derived from the Thomson ISI Web of Science™ (WoS). In this brief communication, the same academics were re‐ranked, using other popular citation databases. It was found that for academics who publish more in computer science forums, theirhwas significantly different due to highly cited papers missed by WoS; consequently, their rank changed substantially. The study was widened to a broader set of UK‐based LIS and IR academics in which results showed similar statistically significant differences. A variant ofh,hmx, was introduced that allowed a ranking of the academics using all citation databases together.

This paper presents an experimental study of users assessing the quality of Google web search results. In particular we look at how users' satisfaction correlates with the effectiveness of Google as quantified by IR measures such as precision and the suite of Cumulative Gain measures (CG, DCG, NDCG). Results indicate strong correlation between users' satisfaction, CG and precision, moderate correlation with DCG, with perhaps surprisingly negligible correlation with NDCG. The reasons for the low correlation with NDCG are examined.

This poster describes a potential problem with a relatively well used measure in Information Retrieval research: Kendall's Tau rank correlation coefficient. The coefficient is best known for its use in determining the similarity of test collections when ranking sets of retrieval runs. Threshold values for the coefficient have been defined and used in a number of published studies in information retrieval. However, this poster presents results showing that basing decisions on such thresholds is not as reliableas has been assumed.

Abstract In this article the authors present Eurovision, a text‐based system for cross‐language (CL) image retrieval. The system is evaluated by multilingual users for two search tasks with the system configured in English and five other languages. To the authors' knowledge, this is the first published set of user experiments for CL image retrieval. They show that (a) it is possible to create a usable multilingual search engine using little knowledge of any language other than English, (b) categorizing images assists the user's search, and (c) there are differences in the way users search between the proposed search tasks. Based on the two search tasks and user feedback, they describe important aspects of any CL image retrieval system.

Abstract A novel and complex form of information access is cross‐language information retrieval: searching for texts written in foreign languages based on native language queries. Although the underlying technology for achieving such a search is relatively well understood, the appropriate interface design is not. The authors present three user evaluations undertaken during the iterative design of Clarity, a cross‐language retrieval system for low‐density languages, and shows how the user‐interaction design evolved depending on the results of usability tests. The first test was instrumental to identify weaknesses in both functionalities and interface; the second was run to determine if query translation should be shown or not; the final was a global assessment and focused on user satisfaction criteria. Lessons were learned at every stage of the process leading to a much more informed view of what a cross‐language retrieval system should offer to users.

The effectiveness of information retrieval systems is measured by comparing performance on a common set of queries and documents. Significance tests are often used to evaluate the reliability of such comparisons. Previous work has examined such tests, but produced results with limited application. Other work established an alternative benchmark for significance, but the resulting test was too stringent. In this paper, we revisit the question of how such tests should be used. We find that the t-test is highly reliable (more so than the sign or Wilcoxon test), and is far more reliable than simply showing a large percentage difference in effectiveness measures between IR systems. Our results show that past empirical work on significance tests over-estimated the error of such tests. We also re-consider comparisons between the reliability of precision at rank 10 and mean average precision, arguing that past comparisons did not consider the assessor effort required to compute such measures. This investigation shows that assessor effort would be better spent building test collections with more topics, each assessed in less detail.

In this poster, we report on the effects of pseudo relevance feedback (PRF) for a cross language image retrieval task using a test collection. Typically PRF has been shown to improve retrieval performance in previous CLIR experiments based on average precision at a fixed rank. However our experiments have shown that queries in which no relevant documents are returned also increases. Because query reformulation for cross language is likely to be harder than with monolingual searching, a great deal of user dissatisfaction would be associated with this scenario. We propose that an additional effectiveness measure based on failed queries may better reflect user satisfaction than average precision alone.

The larger amount of information on the Web is stored in document databases and is not indexed by general-purpose search engines (i.e., Google and Yahoo). Such information is \ndynamically generated through querying databases — which are \nreferred to as Hidden Web databases. Documents returned in \nresponse to a user query are typically presented using templategenerated Web pages. This paper proposes a novel approach that identifies Web page templates by analysing the textual contents and the adjacent tag structures of a document in order to extract query-related data. Preliminary results demonstrate that our approach effectively detects templates and retrieves data with high recall and precision.

Forming test collection relevance judgments from the pooled output of multiple retrieval systems has become the standard process for creating resources such as the TREC, CLEF, and NTCIR test collections. This paper presents a series of experiments examining three different ways of building test collections where no system pooling is used. First, a collection formation technique combining manual feedback and multiple systems is adapted to work with a single retrieval system. Second, an existing method based on pooling the output of multiple manual searches is re-examined: testing a wider range of searchers and retrieval systems than has been examined before. Third, a new approach is explored where the ranked output of a single automatic search on a single retrieval system is assessed for relevance: no pooling whatsoever. Using established techniques for evaluating the quality of relevance judgments, in all three cases, test collections are formed that are as good as TREC.

In this paper, we describe research which could lead to a novel approach to gathering an overview of a document in a foreign language. The research explores how much of the meaning of a document could be represented using images by researching the ability of subjects to derive the search term that might have been used to return a set of images from an image library. The Google image search engine was used to retrieve the images for this experiment, which uses English throughout. The results were analysed with respect to a previous paper [1] exploring ability to recognise concrete objects in hierarchies. It was found that there is a tendency to use one particular level of categorization.

Abstract This report presents a case study of the development of an interface for a novel and complex form of document retrieval: searching for texts written in foreign languages based on native language queries. Although the underlying technology for achieving such a search is relatively well understood, the appropriate interface design is not. A study involving users from the beginning of the design process is described, and it covers initial examination of user needs and tasks, preliminary design and testing of interface components, building, testing, and refining the interface, and, finally, conducting usability tests of the system. Lessons are learned at every stage of the process, leading to a much more informed view of how such an interface should be built.

Most cross language information retrieval research concentrates on language pairs for which direct, rich, and often multiple translation resources already exist. However, for most language pairs, translation via an intermediate language is necessary. Two distinct methods for dealing with the additional ambiguity introduced by the extra translation step have been proposed and individually, shown to improve retrieval effectiveness. Two previous works indicated that in combination, the methods were ineffective. This paper provides strong empirical evidence that the methods can be combined to produce consistent and often significant improvements in retrieval effectiveness. The improvement is shown across a number of different intermediate languages and test collections.

In this poster, we describe an experiment exploring the effectiveness of a pen based text input device for use in query construction. Standard TREC queries were written, recognised, and subsequently retrieved upon. Comparisons between retrieval effectiveness based on the recognised writing and a typed text baseline were made. On average, effectiveness was 75% of the baseline. Other statistics on the quality and nature of recognition are also reported.

This paper reports on the user-centered design methodology and techniques used for the elicitation of user requirements and how these requirements informed the first phase of the user interface design for a Cross-Language Information Retrieval System. We describe a set of factors involved in analysis of the data collected and, finally discuss the implications for user interface design based on the findings.

Our demonstration shows the hierarchy system working on a locally run search engine. Hierarchies are dynamically generated from the retrieved documents, and visualised on the menus. When a user selects a term from the hierarchy, the documents linked to the term are listed, and the term is then added to the initial query to rerun a search. Through the demonstration we illustrate how hierarchical presentation of expansion terms is achieved, and how our approach supports users to articulate their information needs using the hierarchy.

No abstract available.

This poster describes initial work exploring a relatively unexamined area of data fusion: fusing the results of retrieval systems whose collections have no overlap between them. Many of the effective meta-search/data fusion strategies gain much of their success from exploiting document overlap across the source systems being merged. When the intersection of the collections is the empty set, the strategies generally degrade to a simpler form. In order to address such situations, two strategies were examined: re-ranking of merged results using a locally run search on the text fragments returned by the source search engines; and re-ranking based on cross document similarity, again using text fragments presented in the retrieved list. Results, from experiments, which go beyond previous work, indicate that both strategies improve fusion effectiveness.

Most approaches to cross language information retrieval assume that resources providing a direct translation between the query and document languages exist. This paper presents research examining the situation where such an assumption is false. Here, an intermediate (or pivot) language provides a means of transitive translation of the query language to that of the document via the pivot, at the cost, however, of introducing much error. The paper reports the novel approach of translating in parallel across multiple intermediate languages and fusing the results. Such a technique removes the error, raising the effectiveness of the tested retrieval system, up to and possibly above the level expected, had a direct translation route existed. Across a number of retrieval situations and combinations of languages, the approach proves to be highly effective.

This paper presents a system that retrieves descriptive phrases of proper nouns from free text. Sentences holding the specified noun are ranked using a technique based on pattern matching, word counting, and sentence location. No domain specific knowledge is used. Experiments show the system able to rank highly those sentences that contain phrases describing or defining the query noun. In contrast to existing methods, this system does not use parsing techniques but still achieves high levels of accuracy. From the results of a large-scale experiment, it is speculated that the success of this simpler method is due to the high quantities of free text being searched. Parallels between this work and recent findings in the very large corpus track of TREC are drawn. Keywords Information retrieval, descriptive phrase, large corpora. 1. INTRODUCTION The opportunities to use online text databases for the mining of valuable information are great. As these stores increase in size, the possibil...

Although always present in text, word sense ambiguity only recently became regarded as a problem to information retrieval which was potentially solvable. The growth of interest in word senses resulted from new directions taken in disambiguation research. This paper first outlines this research and surveys the resulting efforts in information retrieval. Although the majority of attempts to improve retrieval effectiveness were unsuccessful, much was learnt from the research. Most notably a notion of under what circumstance disambiguation may prove of use to retrieval.

This paper presents a means of automatically deriving a hierarchical organization of concepts from a set of documents without use of training data or standard clustering techniques.Instead, salient words and phrases extracted from the documents are organized hierarchically using a type of co-occurrence known as subsumption.The resulting structure is displayed as a series of hierarchical menus.When generated from a set of retrieved documents, a user browsing the menus is provided with a detailed overview of their content in a manner distinct from existing overview and summarization techniques.The methods used to build the structure are simple, but appear to be effective: a smallscale user study reveals that the generated hierarchy possesses properties expected of such a structure in that general terms are placed at the top levels leading to related and more specific terms below.The formation and presentation of the hierarchy is described along with the user study and some other informal evaluations.

We present an analysis of word senses that provides a fresh insight into the impact of word ambiguity on retrieval effectiveness with potential broader implications for other processes of information retrieval. Using a methodology of forming artifically ambiguous words, known as pseudowords, and through reference to other researchers' work, the analysis illustrates that the distribution of the frequency of occurrance of the senses of a word plays a strong role in ambiguity's impact of effectiveness. Further investigation shows that this analysis may also be applicable to other processes of retrieval, such as Cross Language Information Retrieval, query expansion, retrieval of OCR'ed texts, and stemming. The analysis appears to provide a means of explaining, at least in part, reasons for the processes' impact (or lack of it) on effectiveness.

This paper describes a set of experimental results produced from the TIPSTER SUMMAC initiative on user directed summaries: document summaries generated in the context of an information need expressed as a query. The summarizer that was evaluated was based on a set of existing statistical techniques that had been applied successfully to the INQUERY retrieval system. The techniques proved to have a wider utility, however, as the summarizer was one of the better performing systems in the SUMMAC evaluation. The design of this summarizer is presented with a range of evaluations: both those provided by SUMMAC as well as a set of preliminary, more informal, evaluations that examined additional aspects of the summaries. Amongst other conclusions, the results reveal that users can judge the relevance of documents from their summary almost as accurately as if they had had access to the document's full text. 1.1 Keywords User directed summarization, information retrieval, consistency of relevant judgements. 2.

This paper presents an investigation into the utility of document summarisation in the context of information retrieval, more specifically in the application of so called query biased (or user directed)summaries: summaries customised to reflect the information need expressed in a query.Employed in the retrieved document list displayed after a retrieval took place, the summaries' utility was evaluated in a task-based environment by measuring users' speed and accuracy in identifying relevant documents.This was compared to the performance achieved when users were presented with the more typical output of an IR system: a static predefined summary composed of the title and first few sentences of retrieved documents.The results from the evaluation indicate that the use of query biased summaries significantly improves both the accuracy and speed of user relevance judgements.

This paper presents a model for retrieval of images from a large World Wide Web based collection.Rather than considering complex visual recognition algorithms, the model presented is based on combining evidence of the text content and hypertext structure of the Web.The paper shows that certain types of query are amply served by this form of representation.It also presents a novel means of gathering relevance judgments.