Ranking plays a central role in connecting users and providers in Information Retrieval (IR) systems, making provider-side fairness an important challenge. While recent research has begun to address fairness in ranking, most existing approaches adopt an equality-based perspective, aiming to ensure that providers with similar content receive similar exposure. However, it overlooks the diverse needs of real-world providers, whose utility from ranking may depend not only on exposure but also on outcomes like sales or engagement. Consequently, exposure-based fairness may not accurately capture the true utility perceived by different providers with varying priorities. To this end, we introduce an equity-oriented fairness framework that explicitly models each provider's preferences over key outcomes such as exposure and sales, thus evaluating whether a ranking algorithm can fulfill these individualized goals while maintaining overall fairness across providers. Based on this framework, we develop EquityRank, a gradient-based algorithm that jointly optimizes user-side effectiveness and provider-side equity. Extensive offline and online simulations demonstrate that EquityRank offers improved trade-offs between effectiveness and fairness and adapts to heterogeneous provider needs.

In recent years, the recommendation content on e-commerce platforms has become increasingly rich -- a single user feed may contain multiple entities, such as selling products, short videos, and content posts. To deal with the multi-entity cross-domain recommendation problem, an intuitive solution is to adopt the shared-network-based architecture for joint training. The underlying idea is to transfer the knowledge from one type of entity (source entity) to another (target entity). However, different from the conventional same-entity cross-domain recommendation, multi-entity knowledge transfer encounters several important challenges: (1) data distributions of the source entity and target entity are naturally different, making the shared-network-based joint training susceptible to the negative transfer issue(2) the corresponding feature schema of each entity is not exactly aligned (e.g., price is an essential feature for selling product while missing for content posts), making the existing approaches no longer appropriate. Recent researchers have also experimented with the pre-training and fine-tuning paradigm. Again, they only take into account the scenarios with the same entity type and feature schemas, which is inappropriate in our case. To this end, we design a pre-training & fine-tuning based Multi-entity Knowledge Transfer framework called MKT. MKT utilizes a multi-entity pre-training module to extract transferable knowledge across different entities. In particular, a feature alignment module is first applied to scale and align different feature schemas. Afterward, a couple of knowledge extractors are employed to extract the common and independent knowledge. In the end, the extracted common knowledge is adopted for target entity model training. Through extensive offline and online experiments on public and industrial datasets, we demonstrated the superiority of MKT over multiple State-Of-The-Art methods. MKT has also been deployed for content post recommendations on our production system.

Most recommendation systems typically follow a product-based paradigm utilizing user-product interactions to identify the most engaging items for users. However, this product-based paradigm has notable drawbacks for Xianyu~. Xianyu is China's largest online C2C e-commerce platform where a large portion of the product are post by individual sellers. Most of the product on Xianyu posted from individual sellers often have limited stock available for distribution, and once the product is sold, it's no longer available for distribution. This result in most items distributed product on Xianyu having relatively few interactions, affecting the effectiveness of traditional recommendation depending on accumulating user-item interactions. To address these issues, we introduce IU4Rec, an Interest Unit-based two-stage Recommendation system framework. We first group products into clusters based on attributes such as category, image, and semantics. These IUs are then integrated into the Recommendation system, delivering both product and technological innovations. IU4Rec begins by grouping products into clusters based on attributes such as category, image, and semantics, forming Interest Units (IUs). Then we redesign the recommendation process into two stages. In the first stage, the focus is on recommend these Interest Units, capturing broad-level interests. In the second stage, it guides users to find the best option among similar products within the selected Interest Unit. User-IU interactions are incorporated into our ranking models, offering the advantage of more persistent IU behaviors compared to item-specific interactions. This interest unit based recommendation can be beneficial from mitigating the side effect of limited-stock problem, since most interaction can be gathered on interest units which can persist and accumulate over time. Experimental results on the production dataset and online A/B testing demonstrate the effectiveness and superiority of our proposed IU-centric recommendation approach. This study not only advances recommendation technologies but also emphasizes the potential for co-evolution between product innovations and the technologies involved in item supply and distribution.

Federated Recommendation (FedRec) systems are recognized as privacy-preserving solutions for collaboratively training recommender models without sharing users’ private data. However, recent studies have revealed that FedRec systems are vulnerable to interaction-level membership inference attacks. In such attacks, a semi-honest server can employ crafted methods to infer users’ interacted items. In this article, we identify that user preference information is predominantly stored in the user-uploaded parameters rather than in the local parameters after local training. Leveraging this insight, we expose a new interaction vulnerability and introduce the PubPara attack. Our experiments show that PubPara improves the inference performance by at least 40% over existing attacks, while requiring minimal inference time and remaining robust against current defense methods. To safeguard user privacy without compromising recommender performance, we propose MultiVerse, a novel countermeasure. MultiVerse utilizes untrained items outside the user’s local training data to obfuscate the server’s inference of interacted items. It includes a four-step strategy (training, optimization, refinement, and denoising) to achieve robust defense. Extensive experiments on three representative FedRec models (F-NCF, F-LightGCN, and FedRAP) across three real-world datasets validate that MultiVerse significantly degrades the attack’s inference performance to near the level of random guess while maintaining lossless recommender performance.

In machine learning systems, privileged features refer to the features that are available during offline training but inaccessible for online serving. Previous studies have recognized the importance of privileged features and explored ways to tackle online-offline discrepancies. A typical practice is privileged features distillation (PFD): train a teacher model using all features (including privileged ones) and then distill the knowledge from the teacher model using a student model (excluding the privileged features), which is then employed for online serving. In practice, the pointwise cross-entropy loss is often adopted for PFD. However, this loss is insufficient to distill the ranking ability for CTR prediction. First, it does not consider the non-i.i.d. characteristic of the data distribution, i.e., other items on the same page significantly impact the click probability of the candidate item. Second, it fails to consider the relative item order ranked by the teacher model's predictions, which is essential to distill the ranking ability. To address these issues, we first extend the pointwise-based PFD to the listwise-based PFD. We then define the calibration-compatible property of distillation loss and show that commonly used listwise losses do not satisfy this property when employed as distillation loss, thus compromising the model's calibration ability, which is another important measure for CTR prediction. To tackle this dilemma, we propose Calibration-compatible LIstwise Distillation (CLID), which employs carefully-designed listwise distillation loss to achieve better ranking ability than the pointwise-based PFD while preserving the model's calibration ability. We theoretically prove it is calibration-compatible. Extensive experiments on public datasets and a production dataset collected from the display advertising system of Alibaba further demonstrate the effectiveness of CLID.

Click-Through Rate (CTR) prediction serves as a fundamental component in online advertising. A common practice is to train a CTR model on advertisement (ad) impressions with user feedback. Since ad impressions are purposely selected by the model itself, their distribution differs from the inference distribution and thus exhibits sample selection bias (SSB) that affects model performance. Existing studies on SSB mainly employ sample re-weighting techniques which suffer from high variance and poor model calibration. Another line of work relies on costly uniform data that is inadequate to train industrial models. Thus mitigating SSB in industrial models with a uniform-data-free framework is worth exploring. Fortunately, many platforms display mixed results of organic items (i.e., recommendations) and sponsored items (i.e., ads) to users, where impressions of ads and recommendations are selected by different systems but share the same user decision rationales. Based on the above characteristics, we propose to leverage recommendations samples as a free lunch to mitigate SSB for ads CTR model (Rec4Ad). After elaborating data augmentation, Rec4Ad learns disentangled representations with alignment and decorrelation modules for enhancement. When deployed in Taobao display advertising system, Rec4Ad achieves substantial gains in key business metrics, with a lift of up to +6.6% CTR and +2.9% RPM.

As the self-attention mechanism offers powerful capabilities for capturing sequential relationships, it has become increasingly popular to use it for modeling user behavior sequences in recommender systems. However, the self-attention mechanism has a quadratic computational complexity of O(n^2), as it conducts interactions among all item pairs in the sequence. This can lead to expensive model training and slow inference speeds, which may hinder practical deployment. To this end, we pursue to develop alternative approaches to improve the efficiency of the self-attention mechanism. We observe that the attention scores calculated from each item interacting with other items (including itself) are sparse, indicating that there are limited valuable item pairs (with non-zero attention weight) that contribute to the final output. This motivates us to develop effective strategies for discerning valuable items and computing attention scores solely for these items, thereby minimizing the consumption of unnecessary computations. Herein, we present a novel Progressive Sampling-based Self-Attention (PS-SA) mechanism, which utilizes a learnable progressive sampling strategy to identify the most valuable items. Subsequently, we solely utilize these selected items to produce the final output. Experiments on academic and production datasets demonstrate PS-SA could still achieve promising results while reducing computational costs. It is notable that we have successfully deployed it on Alibaba display advertising system, resulting in a 2.6% CTR and 1.3% RPM increase.

Cascading architecture has been widely adopted in large-scale advertising systems to balance efficiency and effectiveness. In this architecture, the pre-ranking model is expected to be a lightweight approximation of the ranking model, which handles more candidates with strict latency requirements. Due to the gap in model capacity, the pre-ranking and ranking models usually generate inconsistent ranked results, thus hurting the overall system effectiveness. The paradigm of score alignment is proposed to regularize their raw scores to be consistent. However, it suffers from inevitable alignment errors and error amplification by bids when applied in online advertising. To this end, we introduce a consistency-oriented pre-ranking framework for online advertising, which employs a chunk-based sampling module and a plug-and-play rank alignment module to explicitly optimize consistency of ECPM-ranked results. A ΔNDCG-based weighting mechanism is adopted to better distinguish the importance of inter-chunk samples in optimization. Both online and offline experiments have validated the superiority of our framework. When deployed in Taobao display advertising system, it achieves an improvement of up to +12.3% CTR and +5.6% RPM.

Conversion rate (CVR) prediction is an essential task for e-commerce platforms. However, refunds frequently occur after conversion in online shopping systems, which drives us to pay attention to effective conversion for building healthier services. This paper defines the probability of item purchasing without any subsequent refund as an effective conversion rate (ECVR). A simple paradigm for ECVR prediction is to decompose it into two sub-tasks: CVR prediction and post-conversion refund rate (RFR) prediction. However, RFR prediction suffers from data sparsity (DS) and sample selection bias (SSB) issues, as refund behaviors are only available after user purchase. Furthermore, there is delayed feedback in both sequentially dependent conversion and refund events, named cascade delayed feedback (CDF). Previous studies mainly focus on tackling DS and SSB or delayed feedback for a single event. To jointly tackle these issues in ECVR prediction, we propose an Entire space CAscade Delayed feedback modeling (ECAD) method. Specifically, ECAD deals with DS and SSB by constructing two tasks including CVR and conversion&refund rate (CVRFR) predictions using the entire space modeling framework. In addition, it carefully schedules auxiliary tasks to leverage both conversion and refund time within data to alleviate CDF. Experiments on the offline industrial dataset and online A/B testing demonstrate the effectiveness of ECAD. ECAD has been deployed in the Xianyu recommender system of Alibaba, contributing to a significant improvement of ECVR.

Conversion rate (CVR) prediction is one of the core components in online recommender systems, and various approaches have been proposed to obtain accurate and well-calibrated CVR estimation. However, we observe that a well-trained CVR prediction model often performs sub-optimally during sales promotions. This can be largely ascribed to the problem of the data distribution shift, in which the conventional methods no longer work. To this end, we seek to develop alternative modeling techniques for CVR prediction. Observing similar purchase patterns across different promotions, we propose reusing the historical promotion data to capture the promotional conversion patterns. Herein, we propose a novel Historical Data Reuse (HDR) approach that first retrieves historically similar promotion data and then fine-tunes the CVR prediction model with the acquired data for better adaptation to the promotion mode. HDR consists of three components: an automated data retrieval module that seeks similar data from historical promotions, a distribution shift correction module that re-weights the retrieved data for better aligning with the target promotion, and a TransBlock module that quickly fine-tunes the original model for better adaptation to the promotion mode. Experiments conducted with real-world data demonstrate the effectiveness of HDR, as it improves both ranking and calibration metrics to a large extent. HDR has also been deployed on the display advertising system in Alibaba, bringing a lift of 9% RPM and 16% CVR during Double 11 Sales in 2022.

Despite the development of ranking optimization techniques, pointwise loss remains the dominating approach for click-through rate prediction. It can be attributed to the calibration ability of the pointwise loss since the prediction can be viewed as the click probability. In practice, a CTR prediction model is also commonly assessed with the ranking ability. To optimize the ranking ability, ranking loss (e.g., pairwise or listwise loss) can be adopted as they usually achieve better rankings than pointwise loss. Previous studies have experimented with a direct combination of the two losses to obtain the benefit from both losses and observed an improved performance. However, previous studies break the meaning of output logit as the click-through rate, which may lead to sub-optimal solutions. To address this issue, we propose an approach that can Jointly optimize the Ranking and Calibration abilities (JRC for short). JRC improves the ranking ability by contrasting the logit value for the sample with different labels and constrains the predicted probability to be a function of the logit subtraction. We further show that JRC consolidates the interpretation of logits, where the logits model the joint distribution. With such an interpretation, we prove that JRC approximately optimizes the contextualized hybrid discriminative-generative objective. Experiments on public and industrial datasets and online A/B testing show that our approach improves both ranking and calibration abilities. Since May 2022, JRC has been deployed on the display advertising platform of Alibaba and has obtained significant performance improvements.

An industrial recommender system generally presents a hybrid list that contains results from multiple subsystems. In practice, each subsystem is optimized with its own feedback data to avoid the disturbance among different subsystems. However, we argue that such data usage may lead to sub-optimal online performance because of thedata sparsity. To alleviate this issue, we propose to extract knowledge from thesuper-domain that contains web-scale and long-time impression data, and further assist the online recommendation task (downstream task). To this end, we propose a novel industrial KnowlEdge Extraction and Plugging (KEEP) framework, which is a two-stage framework that consists of 1) a supervised pre-training knowledge extraction module on super-domain, and 2) a plug-in network that incorporates the extracted knowledge into the downstream model. This makes it friendly for incremental training of online recommendation. Moreover, we design an efficient empirical approach for KEEP and introduce our hands-on experience during the implementation of KEEP in a large-scale industrial system. Experiments conducted on two real-world datasets demonstrate that KEEP can achieve promising results. It is notable that KEEP has also been deployed on the display advertising system in Alibaba, bringing a lift of +5.4% CTR and +4.7% RPM.

Deep learning techniques have been applied widely in industrial recommendation systems. However, far less attention has been paid on the overfitting problem of models in recommendation systems, which, on the contrary, is recognized as a critical issue for deep neural networks. In the context of Click-Through Rate (CTR) prediction, we observe an interesting one-epoch overfitting problem: the model performance exhibits a dramatic degradation at the beginning of the second epoch. Such a phenomenon has been witnessed widely in real-world applications of CTR models. Thereby, the best performance is usually achieved by training with only one epoch. To understand the underlying factors behind the one-epoch phenomenon, we conduct extensive experiments on the production data set collected from the display advertising system of Alibaba. The results show that the model structure, the optimization algorithm with a fast convergence rate, and the feature sparsity are closely related to the one-epoch phenomenon. We also provide a likely hypothesis for explaining such a phenomenon and conduct a set of proof-of-concept experiments. We hope this work can shed light on the future research on training more epochs for better performance.

Exploration-Exploitation (E& E) algorithms are commonly adopted to deal with the feedback-loop issue in large-scale online recommender systems. Most of existing studies believe that high uncertainty can be a good indicator of potential reward, and thus primarily focus on the estimation of model uncertainty. We argue that such an approach overlooks the subsequent effect of exploration on model training. From the perspective of online learning, the adoption of an exploration strategy would also affect the collecting of training data, which further influences model learning. To understand the interaction between exploration and training, we design a Pseudo-Exploration module that simulates the model updating process after a certain item is explored and the corresponding feedback is received. We further show that such a process is equivalent to adding an adversarial perturbation to the model input, and thereby name our proposed approach as an the Adversarial Gradient Driven Exploration (AGE). For production deployment, we propose a dynamic gating unit to pre-determine the utility of an exploration. This enables us to utilize the limited amount of resources for exploration, and avoid wasting pageview resources on ineffective exploration. The effectiveness of AGE was firstly examined through an extensive number of ablation studies on an academic dataset. Meanwhile, AGE has also been deployed to one of the world-leading display advertising platforms, and we observe significant improvements on various top-line evaluation metrics.

Learning to Rank, a central problem in information retrieval, is a class of machine learning algorithms that formulate ranking as an optimization task. The objective is to learn a function that produces an ordering of a set of documents in such a way that the utility of the entire ordered list is maximized. Learning-to-rank methods do so by learning a function that computes a score for each document in the set. A ranked list is then compiled by sorting documents according to their scores. While such a deterministic mapping of scores to permutations makes sense during inference where stability of ranked lists is required, we argue that its greedy nature during training leads to less robust models. This is particularly problematic when the loss function under optimization---in agreement with ranking metrics---largely penalizes incorrect rankings and does not take into account the distribution of raw scores. In this work, we present a stochastic framework where, instead of a deterministic derivation of permutations from raw scores, permutations are sampled from a distribution defined by raw scores. Our proposed sampling method is differentiable and works well with gradient descent optimizers. We analytically study our proposed method and demonstrate when and why it leads to model robustness. We also show empirically, through experiments on publicly available learning-to-rank datasets, that the application of our proposed method to a class of ranking loss functions leads to significant model quality improvements.

The rapid growth of commercial web content has driven the development of shopping search services to help users find product offers. Due to the dynamic nature of commercial content, an effective recrawl policy is a key component in a shopping search service; it ensures that users have access to the up-to-date product details. Most of the existing strategies either relied on simple heuristics, or overlooked the resource budgets. To address this, Azar et al. [5] recently proposed an optimization strategy LambdaCrawl aiming to maximize content freshness within a given resource budget. In this paper, we demonstrate that the effectiveness of LambdaCrawl is governed in large part by how well future content change rate can be estimated. By adopting the state-of-the-art deep learning models for change rate prediction, we obtain a substantial increase of content freshness over the common LambdaCrawl implementation with change rate estimated from the past history. Moreover, we demonstrate that while LambdaCrawl is a significant advancement upon existing recrawl strategies, it can be further improved upon by a unified multi-strategy recrawl policy. To this end, we adopt the K-armed adversarial bandits algorithm that can provably optimize the overall freshness by combining multiple strategies. Empirical results over a large-scale production dataset confirm its superiority to LambdaCrawl, especially under tight resource budgets.

Web crawlers spend significant resources to maintain freshness of their crawled data. This paper describes the optimization of resources to ensure that product prices shown in ads in a context of a shopping sponsored search service are synchronized with current merchant prices. We are able to use the predictability of price changes to build a machine learned system leading to considerable resource savings for both the merchants and the crawler. We describe our solution to technical challenges due to partial observability of price history, feedback loops arising from applying machine learned models, and offers in cold start state. Empirical evaluation over large-scale product crawl data demonstrates the effectiveness of our model and confirms its robustness towards unseen data. We argue that our approach can be applicable in more general data pull settings.

In this study, we examined what sources laypeople would select (i.e., visit and adopt) to resolve their health-related information needs, and how different health conditions affect the selection. Twenty-four college students participated in this user study, where they were asked to search for two separate health issues respectively: multiple sclerosis and weight loss. The search logs were collected and analyzed afterwards. We classify the online information sources on both website level and webpage level, and a webpage classification scheme based on genre is proposed. Results suggest that users» selection of sources depends on different types of health issues in terms of urgency and complexity. Health-specific webpage is a popular source and highly adopted for both tasks, but it is particularly helpful for urgent and complex health conditions. Search engines could facilitate users to navigate among scattered health information and support concerns regarding common health issues.

Recent research shows that Collaborative Information Retrieval (CIR), in which two or more users collaborate on the same search task, has become increasingly popular. The presence of both search and collaboration behaviors makes CIR a complex search format, which further drives a critical need to understand CIR's search context. The contextual support for CIR should consider search contexts derived from both team members' search histories (including users' own search histories and partners' search histories) and their explicit collaboration (e.g., chatting). As it stands, existing studies on contextual search support only focus on Individual Information Retrieval (IIR) and only utilize individuals' own search histories. In this paper, we examine the unique search contexts (e.g., partners' search histories and team collaboration histories) in CIR. Based on a user study data collection with 54 participants, we find that compared to the use of individuals' own search histories, CIR contextual support is more effective when utilizing partners' search histories and teams' collaboration behaviors. More interestingly, though the explicit communication information (i.e., chat content) often involves massive noisy information, involving such noise does not affect the ranking of relevant documents since it also does not appear in relevant documents.

When using search engines, people often issue multiple related queries to accomplish a complex search task. A simple query-task structure may not fully capture the complexity of query relations since people may divide a task into multiple subtasks. As a result, this paper applies a three-level hierarchical structure with query, goal and mission - a mission includes several goals, and a goal consists of multiple queries. Particularly, we focus on analyzing query-goal-mission structure for mobile web search because of its increasing popularity and lack of investigation in the literature. This study has three main contributions: (1) we study the query-goal-mission structure for mobile web search, which was not studied before. (2) We identify several differences between mobile and desktop search patterns in terms of goal/mission length, duration and interleaving. (3) We demonstrate that the query-goal-mission structure can be applied to design better user satisfaction metrics. Specifically, goal-based search success rate and mission-based abandonment rate are better aligned with users' long-term engagement than query and session based metrics.

Although the volume of online educational resources has dramatically increased in recent years, many of these resources are isolated and distributed in diverse websites and databases. This hinders the discovery and overall usage of online educational resources. By using linking between related subsections of online textbooks as a testbed, this paper explores multiple knowledge-based content linking algorithms for connecting online educational resources. We focus on examining semantic-based methods for identifying important knowledge components in textbooks and their usefulness in linking book subsections. To overcome the data sparsity in representing textbook content, we evaluated the utility of external corpuses, such as more textbooks or other online educational resources in the same domain. Our results show that semantic modeling can be integrated with a term-based approach for additional performance improvement, and that using extra textbooks significantly benefits semantic modeling. Similar results are obtained when we applied the same approach to other domains.

Parsing the semantic structure of a web page is a key component of web information extraction. Successful extraction algorithms usually require large-scale training and evaluation datasets, which are difficult to acquire. Recently, crowdsourcing has proven to be an effective method of collecting large-scale training data in domains that do not require much domain knowledge. For more complex domains, researchers have proposed sophisticated quality control mechanisms to replicate tasks in parallel or sequential ways and then aggregate responses from multiple workers. Conventional annotation integration methods often put more trust in the workers with high historical performance; thus, they are called performance-based methods. Recently, Rzeszotarski and Kittur have demonstrated that behavioral features are also highly correlated with annotation quality in several crowdsourcing applications. In this article, we present a new crowdsourcing system, called Wernicke, to provide annotations for web information extraction. Wernicke collects a wide set of behavioral features and, based on these features, predicts annotation quality for a challenging task domain: annotating web page structure. We evaluate the effectiveness of quality control using behavioral features through a case study where 32 workers annotate 200 Q&A web pages from five popular websites. In doing so, we discover several things: (1) Many behavioral features are significant predictors for crowdsourcing quality. (2) The behavioral-feature-based method outperforms performance-based methods in recall prediction, while performing equally with precision prediction. In addition, using behavioral features is less vulnerable to the cold-start problem, and the corresponding prediction model is more generalizable for predicting recall than precision for cross-website quality analysis. (3) One can effectively combine workers’ behavioral information and historical performance information to further reduce prediction errors.

Mobile devices enable people to look for information at the moment when their information needs are triggered. While experiencing complex information needs that require multiple search sessions, users may utilize desktop computers to fulfill information needs started on mobile devices. Under the context of mobile-to-desktop web search, this article analyzes users’ behavioral patterns and compares them to the patterns in desktop-to-desktop web search. Then, we examine several approaches of using Mobile Touch Interactions (MTIs) to infer relevant content so that such content can be used for supporting subsequent search queries on desktop computers. The experimental data used in this article was collected through a user study involving 24 participants and six properly designed cross-device web search tasks. Our experimental results show that (1) users’ mobile-to-desktop search behaviors do significantly differ from desktop-to-desktop search behaviors in terms of information exploration, sense-making and repeated behaviors. (2) MTIs can be employed to predict the relevance of click-through documents, but applying document-level relevant content based on the predicted relevance does not improve search performance. (3) MTIs can also be used to identify the relevant text chunks at a fine-grained subdocument level. Such relevant information can achieve better search performance than the document-level relevant content. In addition, such subdocument relevant information can be combined with document-level relevance to further improve the search performance. However, the effectiveness of these methods relies on the sufficiency of click-through documents. (4) MTIs can also be obtained from the Search Engine Results Pages (SERPs). The subdocument feedbacks inferred from this set of MTIs even outperform the MTI-based subdocument feedback from the click-through documents.

People search is an active research topic in recent years. Related works includes expert finding, collaborator recommendation, link prediction and social matching. However, the diverse objectives and exploratory nature of those tasks make it difficult to develop a flexible method for people search that works for every task. In this project, we developed PeopleExplorer, an interactive people search system to support exploratory search tasks when looking for people. In the system, users could specify their task objectives by selecting and adjusting key criteria. Three criteria were considered: the content relevance, the candidate authoritativeness and the social similarity between the user and the candidates. This project represents a first attempt to add transparency to exploratory people search, and to give users full control over the search process. The system was evaluated through an experiment with 24 participants undertaking four different tasks. The results show that with comparable time and effort, users of our system performed significantly better in their people search tasks than those using the baseline system. Users of our system also exhibited many unique behaviors in query reformulation and candidate selection. We found that users' general perceptions about three criteria varied during different tasks, which confirms our assumptions regarding modeling task difference and user variance in people search systems.

We propose a method to dynamically estimate the utility of documents in a search session by modeling the users' browsing behaviors and novelty. The method can be applied to evaluate query reformulations in a search session.

This paper presents a user study aiming to investigate the query reformulation in collaborative Web search. 7 pairs of participants were recruited and each pair worked as a team on two collaborative exploratory Web search tasks. Through the log analysis, we compared possible sources for participants to draw query terms from. The results show that both search and collaborative actions are possible resources for new query terms. Traditional resources for query expansion such as previous search histories and relevant documents are still important resources for new query terms. The content in chat and workspace generated by participants themselves seems more likely to be the resource for new query terms than that of their partners. Task types also affect the influences on query reformulations. For the academic task, previously saved relevance documents are the most important resources for new query terms while chat histories are the most important resources for the leisure task.

Current search systems are designed to find relevant articles, especially topically relevant ones, but the notion of relevance largely depends on search tasks. We study the specific task that scientists are searching for worth-reading articles beneficial for their research. Our study finds: users' perception of relevance and preference of reading are only moderately correlated; current systems can effectively find readings that are highly relevant to the topic, but 36% of the worth-reading articles are only marginally relevant or even non-relevant. Our system can effectively find those worth-reading but marginally relevant or non-relevant articles by taking advantages of scientists' recommendations in social websites.