Search

Due to the existence of large numbers of legacy documents (such as old books and newspapers), improving retrieval effectiveness for OCR'ed documents continues to be an important problem. This article compares the effect of OCR error correction with and without language modeling and the effect of query garbling with weighted structured queries on the retrieval of OCR degraded Arabic documents. The results suggest that moderate error correction does not yield statistically significant improvement in retrieval effectiveness when indexing and searching using n-grams. Also, reversing error correction models to perform query garbling in conjunction with weighted structured queries yields improved retrieval effectiveness. Lastly, using very good error correction that utilizes language modeling yields the best improvement in retrieval effectiveness. [ABSTRACT FROM AUTHOR]

Language modeling is essential in Natural Language Processing and Information Retrieval related tasks. After the statistical language models, Quantum Language Model (QLM) has been proposed to unify both single words and compound terms in the same probability space without extending term space exponentially. Although QLM achieved good performance in ad hoc retrieval, it still has two major limitations: (1) QLM cannot make use of supervised information, mainly due to the iterative and non-differentiable estimation of the density matrix, which represents both queries and documents in QLM. (2) QLM assumes the exchangeability of words or word dependencies, neglecting the order or position information of words. This article aims to generalize QLM and make it applicable to more complicated matching tasks (e.g., Question Answering) beyond ad hoc retrieval. We propose a complex-valued neural network-based QLM solution called C-NNQLM to employ an end-to-end approach to build and train density matrices in a lightweight and differentiable manner, and it can therefore make use of external well-trained word vectors and supervised labels. Furthermore, C-NNQLM adopts complex-valued word vectors whose phase vectors can directly encode the order (or position) information of words. Note that complex numbers are also essential in the quantum theory. We show that the real-valued NNQLM (R-NNQLM) is a special case of C-NNQLM. The experimental results on the QA task show that both R-NNQLM and C-NNQLM achieve much better performance than the vanilla QLM, and C-NNQLM’s performance is on par with state-of-the-art neural network models. We also evaluate the proposed C-NNQLM on text classification and document retrieval tasks. The results on most datasets show that the C-NNQLM can outperform R-NNQLM, which demonstrates the usefulness of the complex representation for words and sentences in C-NNQLM. [ABSTRACT FROM AUTHOR]

The language-modeling approach to information retrieval provides an effective statistical framework for tackling various problems and often achieves impressive empirical performance. However, most previous work on language models for information retrieval focused on document-specific characteristics, and therefore did not take into account the structure of the surrounding corpus, a potentially rich source of additional information. We propose a novel algorithmic framework in which information provided by document-based language models is enhanced by the incorporation of information drawn from clusters of similar documents. Using this framework, we develop a suite of new algorithms. Even the simplest typically outperforms the standard language-modeling approach in terms of mean average precision (MAP) and recall, and our new interpolation algorithm posts statistically significant performance improvements for both metrics over all six corpora tested. An important aspect of our work is the way we model corpus structure. In contrast to most previous work on cluster-based retrieval that partitions the corpus, we demonstrate the effectiveness of a simple strategy based on a nearest-neighbors approach that produces overlapping clusters. [ABSTRACT FROM AUTHOR]

The article investigates how multilingual dialogue systems can be enhanced by leveraging information from other languages through a collaborative framework of expert agents. Topics include the impact of data sufficiency and language characteristics on system performance, the design of a mixture-of-languages routing framework to optimize multilingual task-oriented dialogues, and the challenges of acquiring multilingual datasets and achieving effective language transfer.

The article presents a dual contrastive learning model, Dual-C that enhances cross-domain named entity recognition (NER) by refining token-level and sentence-level representations to improve entity span detection (ESD) and entity type classification (ETC). It highlights the challenges faced by existing NER methods in distinguishing entities from non-entities and the decomposition of NER tasks into subtasks for better performance.

The article focuses on elastic language model ElasticMl that dynamically adjusts its latency-performance tradeoff based on varying request volumes, particularly for information retrieval tasks. Topics include the limitations of static tradeoffs in conventional language models, the design of an elastic structure and optimization method for enabling compute elasticity, and the adaptation of ElasticLM into specialized models for dense retrieval and reranking.

The article focuses on the Self-Distilled Learning to Rank (SDLR) framework, which is designed to improve the generalizability of ad hoc retrieval models through a self-distillation process. Topics include the methodology of assigning confidence weights to training samples based on feature distributions, the comparison of SDLR's performance against state-of-the-art learning to rank models using evaluation metrics and the effectiveness of the student model trained on a portion of data.

The article focuses on enhancing Cross-domain Named Entity Recognition (NER) by recognizing the discrepant transferability between entity detection and type prediction tasks. It introduces a novel paradigm called divide-and-transfer, employing separate functional modules for each task to improve domain transfer effectiveness.

The article focuses on improving generative retrieval (GR) models by introducing a listwise approach that optimizes relevance at the document list level, unlike existing models that use pointwise approaches. It mentions by treating the generation of a ranked document list as a sequence learning process, the proposed method maximizes the generation likelihood of each document given the preceding documents in the list, addressing the sub-optimality of pointwise approaches.

The article focuses on enhancing search result diversification by acknowledging that long documents cover various aspects of a query and proposing to represent documents using multiple passages. It mentions by segmenting documents into passages and evaluating their relevance to the query, the proposed framework improves result diversity compared to existing methods, as demonstrated through experiments on a standard dataset.

The article focuses on advancing sparse representation learning within Information Retrieval, leveraging Pre-trained Language Models to unify term weighting and expansion in a supervised framework, resulting in state-of-the-art performance (SPLADE++). It mentions efficiency improvements are proposed, ensuring latency requirements comparable to traditional keyword-based approaches (Efficient-SPLADE), thereby enhancing the practicality of these methods in real-world applications.

The article focuses on effective dialogue planning for target-oriented dialogue generation, crucial for leading conversations toward predetermined targets. It mentions through the proposed TRIP approach, inspired by decision-making theories, the system plans dialogue paths bidirectionally to generate action-topic pairs, improving planning effectiveness and converging on consistent actions.

The article focuses on improving the efficiency of late-interaction multi-vector models like ColBERT and COIL, which achieve state-of-the-art retrieval effectiveness but suffer from high storage overhead. It mentions through systematic analysis, the study proposes document pruning methods to reduce storage requirements and compares the effectiveness of different pruning techniques, along with leveraging query pruning methods to decrease retrieval latency.

The article focuses on training Neural Ranking Models (NRMs) via weak supervision, using automatically generated datasets from existing ranking models, which significantly reduces the need for manually labeled data. It mentions through an iterative re-labeling process called Generalized Weak Supervision (GWS), weakly supervised models iteratively improve ranking performance without manual labeling, offering four implementations: self-labeling.

The article focuses on enhancing efficiency in modern transformer-based information retrieval models, addressing the limitations posed by the self-attention mechanism for long input, such as in document retrieval, by exploring bag-of-words document representations that encode documents using only a fraction of their characteristic terms. It mentions through experiments on MS MARCO and Robust04 datasets.

The article focuses on enhancing contextual ranking models by proposing data augmentation methods to improve ranking performance effectively and robustly. It mentions by utilizing supervised and unsupervised augmentation schemes, along with contrastive losses adapted for ranking tasks, the study demonstrates significant performance improvements, particularly in sample efficiency and robustness across in-domain and out-of-domain benchmarks.

The article focuses on addressing the inefficiency of dual-encoder-based dense retrieval models in information retrieval (IR) by proposing Fast-Forward indexes, which leverage semantic matching capabilities for efficient re-ranking. By enhancing computational efficiency and optimizing memory footprint, Fast-Forward indexes achieve low latency and competitive results in IR tasks without requiring hardware acceleration like GPUs.

The article focuses on improving Query-by-Document (QBD) retrieval, a challenging task in information retrieval, by proposing a Re-Ranker based on the Proportional Relevance Score (RPRS) to compute relevance between queries and candidate documents. RPRS significantly outperforms existing models on various datasets, offering high efficiency with a complexity of O(N) and addressing the low-resource training issue in QBD retrieval tasks.

The article delves into predicting users' information needs by scrutinizing their digital activities, employing transformer models. Topics discussed include the impact of comprehensive context data, model enhancement, and performance evaluation against ground truth data. This research aims to extend current practices and achieve superior search interaction support and performance.

The article introduces a novel approach to personalized web search by dynamically modeling user interests based on historical and current queries, improving search result relevance. Topics include challenges with static document representations, the need for dynamic representations, and experimental validation of the proposed model's effectiveness.

The article explores dense text retrieval methods using pretrained language models (PLMs), focusing on leveraging PLM capabilities for semantic matching and relevance modeling. Topics include PLM-based dense retrieval architecture, training strategies, and indexing and integration techniques, providing a comprehensive overview of recent advancements in the field.

Although the seminal proposal to introduce language modeling in information retrieval was based on a multivariate Bernoulli model, the predominant modeling approach is now centered on multinomial models. Language modeling for retrieval based on multivariate Bernoulli distributions is seen inefficient and believed less effective than the multinomial model. In this article, we examine the multivariate Bernoulli model with respect to its successor and examine its role in future retrieval systems. In the context of Bayesian learning, these two modeling approaches are described, contrasted, and compared both theoretically and computationally. We show that the query likelihood following a multivariate Bernoulli distribution introduces interesting retrieval features which may be useful for specific retrieval tasks such as sentence retrieval. Then, we address the efficiency aspect and show that algorithms can be designed to perform retrieval efficiently for multivariate Bernoulli models, before performing an empirical comparison to study the behaviorial aspects of the models. A series of comparisons is then conducted on a number of test collections and retrieval tasks to determine the empirical and practical differences between the different models. Our results indicate that for sentence retrieval the multivariate Bernoulli model can significantly outperform the multinomial model. However, for the other tasks the multinomial model provides consistently better performance (and in most cases significantly so). An analysis of the various retrieval characteristics reveals that the multivariate Bernoulli model tends to promote long documents whose nonquery terms are informative. While this is detrimental to the task of document retrieval (documents tend to contain considerable nonquery content), it is valuable for other tasks such as sentence retrieval, where the retrieved elements are very short and focused. [ABSTRACT FROM AUTHOR]

The article focuses on proposing a sequential prediction framework with Decoupled Progressive Distillation (DePoD), addressing challenges in capturing comprehensive intents for resource allocation in sequential prediction. Topics include sequential prediction, representation learning, and interaction dynamics in modeling intents for improving accuracy-based metrics in various applications such as recommender systems and predictive policing.

The article focuses on proposing an efficient architecture, SMLP4Rec, for sequential recommendations, addressing issues in existing self-attention models and MLP4Rec. Topics include changing the cascading structure to parallel mode, integrating normalization layers for efficiency, and improving training speed and prediction accuracy compared to MLP4Rec.

The article focuses on Information Retrieval Evaluation Measures defined on Axiomatic Models of Preferences (AMPs). Topics include the formal exploration of numeric, metric, and scale properties of effectiveness measures, the identification of join-irreducible elements in AMPs, and the characterization of retrieval measures such as precision, recall, and AP in terms of these elements.

The article focuses on contextualizing and expanding conversational queries without supervision, addressing conversational dependencies and query resolution challenges. Topics include zero-shot unified resolution-retrieval approaches, contextualization of query embeddings using conversation history, and query expansion techniques for improving conversational retrieval.

The article focuses on improving the first-stage retrieval of Point-of-interest (POI) search by pre-training models. Topics include limitations of existing first-stage POI retrieval models, the proposed Geo-Enhanced Dense Retrieval framework, and the use of pre-trained language models to better model semantic match and geographical context information.

The article focuses on developing an opinion-based explainable recommendation model, addressing challenges in understanding the complex relationship between user opinions and ratings and providing personalized explanations. Topics include joint modeling of user-item interactions for preference-aware rating prediction, alleviating sparsity issues in opinion prediction, and validating the model's effectiveness through experiments on popular datasets.

The article proposes a Structured Legal case Retrieval (SLR) framework for legal case retrieval, addressing challenges unique to this task. It introduces internal structural information to handle lengthy legal documents, segmenting them into key components like Facts, Holding, and Decision. It is reported that this segmentation allows the model to process segments separately, preserving essential information.

The article proposes CKGE, a contextualized knowledge graph (KG) embedding approach for explainable talent training course recommendation. It addresses the challenge of providing explainable recommendations by considering different learning motivations. It is reported that CKGE integrates contextualized neighbor semantics and high-order connections as motivation-aware information for effective representation learning of talents and courses.

The article presents STAR-HiT, a model for improved next Point-of-Interest (POI) recommendations in location-based social networks. It utilizes hierarchical encoders to capture spatio-temporal context and identify diverse patterns in user check-in sequences. It is reported that STAR-HiT excels in predicting the next POI, providing superior performance and interpretable recommendations based on experiments with three datasets.

The article introduces PARADE (Passage Representation Aggregation for Document Reranking), focusing on improving document ranking using pre-trained transformer models like BERT and T5. It addresses the limitations of existing models, which process document passages individually due to sequence length constraints. It explores techniques for aggregating relevance signals from document passages, showcasing significant improvements over prior score aggregation methods.

The article focuses on improving the efficiency of coarse-grained response selection in retrieval-based chatbots. It address challenges faced by dense retrieval models, including the inefficiency of multi-turn dialogue history re-computation and the significant storage requirements for offline indexing. It proposes two methods, Hierarchical Dense Retrieval (HDR-BERT) for efficient dialogue history processing and Deep Semantic Hashing (DSH) for compact index storage.

The article focuses on improving search parameter selection for information retrieval systems, emphasizing the need to adapt search configurations to individual queries. Topics include the challenges of selecting optimal search configurations, risk-sensitive criteria for configuration selection, and the empirical evaluation of the proposed approach's effectiveness, particularly in handling diverse query types.

The article focuses on improving the faithfulness and scrutability of model-agnostic explanations in recommender systems. Topics include the development of a model-agnostic explanation framework, a reusable evaluation pipeline for explanation quality, and the use of knowledge distillation to enhance faithfulness and scrutability.

The article focuses on stylized data-to-text generation in the e-commerce domain, introducing a new task where the goal is to generate coherent text for non-linguistic data according to specific styles. Topics include the challenges ensuring the logic of generated text, handling unstructured style references, and mitigating biased training samples.

Spoken term detection (STD) is a key technology for spoken information retrieval. As compared to the conventional speech transcription and keyword spotting, STD is an open-vocabulary task and has to address out-of-vocabulary (OOV) terms. Approaches based on subword units, for example phones, are widely used to solve the OOV issue; however, performance on OOV terms is still substantially inferior to that of in-vocabulary (INV) terms. The performance degradation on OOV terms can be attributed to a multitude of factors. One particular factor we address in this article is the unreliable confidence estimation caused by weak acoustic and language modeling due to the absence of OOV terms in the training corpora. We propose a direct posterior confidence derived from a discriminative model, such as multilayer perceptron (MLP). The new confidence considers a wide-range acoustic context which is usually important for speech recognition and retrieval;moreover, it localizes on detected speech segments and therefore avoids the impact of long-span word context which is usually unreliable for OOV term detection. In this article, we first develop an extensive discussion about the modeling weakness problem associated with OOV terms, and then propose our approach to address this problem based on direct poster confidence. Our experiments carried out on spontaneous and conversational multiparty meeting speech, demonstrate that the proposed technique provides a significant improvement in STD performance as compared to conventional lattice-based confidence, in particular for OOV terms. Furthermore, the new confidence estimation approach is fused with other advanced techniques for OOV treatment, such as stochastic pronunciation modeling and discriminative confidence normalization. This leads to an integrated solution for OOV term detection that results in a large performance improvement. [ABSTRACT FROM AUTHOR]

The article discusses methods to enhance the interpretability of recommendation systems by leveraging pre-trained Transformer models and prompt learning techniques. It explores discrete and continuous prompt learning approaches for incorporating user and item IDs into the models, along with training strategies to improve their performance, achieving superior results on explainable recommendation tasks.

The article focuses on addressing the lack of interpretability in neural document ranking models by introducing a Select-And-Rank paradigm that generates explanations for ranking decisions through sentence selection. It aims to make ranking models interpretable and demonstrates its applicability in various ranking tasks.

The article's focus lies in examining the vulnerability of neural ranking models (NRMs) to adversarial attacks, particularly emphasizing the Word Substitution Ranking Attack (WSRA) task, aiming to enhance a target document's ranking by subtly altering its text. It covers topics such as proposing a novel PRADA method for generating adversarial examples, conducting experiments on web search datasets, and highlighting the importance of identifying NRM vulnerabilities to enhance model robustness.

The article focuses on the examination of learned sparse representations generated by transformers in the context of document-at-a-time (DaaT) and score-at-a-time (SaaT) query evaluation for ranking models. The topics covered include the impact of term weights on query performance, a comparison of DaaT and SaaT approaches in this context, and the introduction of optimizations to enhance retrieval efficiency while maintaining effectiveness.

The Visual Question Answering (VQA) task utilizes both visual image and language analysis to answer a textual question with respect to an image. It has been a popular research topic with an increasing number of real-world applications in the last decade. This paper introduces a novel hierarchical integration of vision and language AliceMind-MMU (ALIbaba’s Collection of Encoder-decoders from Machine IntelligeNce lab of Damo academy - MultiMedia Understanding), which leads to similar or even slightly better results than a human being does on VQA. A hierarchical framework is designed to tackle the practical problems of VQA in a cascade manner including: (1) diverse visual semantics learning for comprehensive image content understanding; (2) enhanced multi-modal pre-training with modality adaptive attention; and (3) a knowledge-guided model integration with three specialized expert modules for the complex VQA task. Treating different types of visual questions with corresponding expertise needed plays an important role in boosting the performance of our VQA architecture up to the human level. An extensive set of experiments and analysis are conducted to demonstrate the effectiveness of the new research work. [ABSTRACT FROM AUTHOR]

Pseudo Relevance Feedback (PRF) is known to improve the effectiveness of bag-of-words retrievers. At the same time, deep language models have been shown to outperform traditional bag-of-words rerankers. However, it is unclear how to integrate PRF directly with emergent deep language models. This article addresses this gap by investigating methods for integrating PRF signals with rerankers and dense retrievers based on deep language models. We consider text-based, vector-based and hybrid PRF approaches and investigate different ways of combining and scoring relevance signals. An extensive empirical evaluation was conducted across four different datasets and two task settings (retrieval and ranking). Text-based PRF results show that the use of PRF had a mixed effect on deep rerankers across different datasets. We found that the best effectiveness was achieved when (i) directly concatenating each PRF passage with the query, searching with the new set of queries, and then aggregating the scores; (ii) using Borda to aggregate scores from PRF runs. Vector-based PRF results show that the use of PRF enhanced the effectiveness of deep rerankers and dense retrievers over several evaluation metrics. We found that higher effectiveness was achieved when (i) the query retains either the majority or the same weight within the PRF mechanism, and (ii) a shallower PRF signal (i.e., a smaller number of top-ranked passages) was employed, rather than a deeper signal. Our vector-based PRF method is computationally efficient; thus, this represents a general PRF method others can use with deep rerankers and dense retrievers. [ABSTRACT FROM AUTHOR]

Recent yearswitnessed several advances in developing multi-goal conversational recommender systems (MGCRS) that can proactively attract users’ interests and naturally lead user-engaged dialogues with multiple conversational goals and diverse topics. Four tasks are often involved in MG-CRS, including Goal Planning, Topic Prediction, Item Recommendation, and Response Generation. Most existing studies address only some of these tasks. To handle the whole problem of MG-CRS, modularized frameworks are adopted where each task is tackled independently without considering their interdependencies. In this work, we propose a novel Unified MultI-goal conversational recommeNDer system (UniMIND). Specifically, we unify these four tasks with different formulations into the same sequence-to-sequence paradigm. Prompt-based learning strategies are investigated to endow the unified model with the capability of multi-task learning. Finally, the overall learning and inference procedure consists of three stages, including multi-task learning, prompt-based tuning, and inference. Experimental results on two MG-CRS benchmarks (DuRecDial and TG-ReDial) show that UniMIND achieves state-of-the-art performance on all tasks with a unified model. Extensive analyses and discussions are provided for shedding some new perspectives for MG-CRS. [ABSTRACT FROM AUTHOR]

Modern recommender systems are trained to predict users’ potential future interactions fromusers’ historical behavior data. During the interaction process, despite the data coming from the user side, recommender systems also generate exposure data to provide users with personalized recommendation slates. Compared with the sparse user behavior data, the system exposure data are much larger in volume since only very few exposed items would be clicked by the user. In addition, user historical behavior data are privacy sensitive and commonly protected with careful access authorization. However, the large volume of recommender exposure data generated by the service provider itself usually receives less attention and could be accessed within a relatively larger scope of various information seekers or even potential adversaries. In this article,we investigate the problem of user behavior data leakage in the field of recommender systems. We show that the privacy-sensitive user past behavior data can be inferred through the modeling of system exposure. In other words, one can infer which items the user has clicked just from the observation of current system exposure for this user. Given the fact that system exposure data could be widely accessed from a relatively larger scope, we believe that user past behavior privacy has a high risk of leakage in recommender systems. More precisely, we conduct an attack model whose input is the current recommended item slate (i.e., system exposure) for the user while the output is the user’s historical behavior. Specifically, we exploit an encoder-decoder structure to construct the attack model and apply different encoding and decoding strategies to verify attack performance. Experimental results on two real-world datasets indicate a great danger of user behavior data leakage. To address the risk, we propose a two-stage privacy-protection mechanism that first selects a subset of items from the exposure slate and then replaces the selected items with uniform or popularity-based exposure. Experimental evaluation reveals a trade-off effect between the recommendation accuracy and the privacy disclosure risk, which is an interesting and important topic for privacy concerns in recommender systems. [ABSTRACT FROM AUTHOR]

On a wide range of natural language processing and information retrieval tasks, transformer-based models, particularly pre-trained language models like BERT, have demonstrated tremendous effectiveness. Due to the quadratic complexity of the self-attention mechanism, however, such models have difficulties processing long documents. Recent works dealing with this issue include truncating long documents, in which case one loses potential relevant information, segmenting them into several passages, which may lead to miss some information and high computational complexity when the number of passages is large, or modifying the self-attention mechanism to make it sparser as in sparse-attention models, at the risk again of missing some information.We follow here a slightly different approach in which one first selects key blocks of a long document by local query-block pre-ranking, and then few blocks are aggregated to form a short document that can be processed by a model such as BERT. Experiments conducted on standard Information Retrieval datasets demonstrate the effectiveness of the proposed approach. [ABSTRACT FROM AUTHOR]

One of the key challenges in sequential recommendation is how to extract and represent user preferences. Traditional methods rely solely on predicting the next item. But user behavior may be driven by complex preferences. Therefore, these methods cannot make accurate recommendations when the available information user behavior is limited. To explore multiple user preferences, we propose a transformer-based sequential recommendation model, named MrTransformer (Multi-preference Transformer). For training MrTransformer, we devise a preference-editing-based self-supervised learning (SSL) mechanism that explores extra supervision signals based on relations with other sequences. The idea is to force the sequential recommendation model to discriminate between common and unique preferences in different sequences of interactions. By doing so, the sequential recommendation model is able to disentangle user preferences into multiple independent preference representations so as to improve user preference extraction and representation. We carry out extensive experiments on five benchmark datasets. MrTransformer with preference editing significantly outperforms state-of-the-art sequential recommendation methods in terms of Recall, MRR, and NDCG. We find that long sequences of interactions from which user preferences are harder to extract and represent benefit most from preference editing. [ABSTRACT FROM AUTHOR]

With the resurgent interest in building open-domain dialogue systems, the dialogue generation task has attracted increasing attention over the past few years. This task is usually formulated as a conditional generation problem, which aims to generate a natural and meaningful response given dialogue contexts and specific constraints, such as persona. And maintaining a consistent persona is essential for the dialogue systems to gain trust from the users. Although tremendous advancements have been brought, traditional persona-based dialogue models are typically trained by leveraging a large number of persona-dense dialogue examples. Yet, such persona-dense training data are expensive to obtain, leading to a limited scale. This work presents a novel approach to learning from limited training examples by regarding consistency understanding as a regularization of response generation. To this end, we propose a novel stack-propagation framework for learning a generation and understanding pipeline. Specifically, the framework stacks a Transformer encoder and two Transformer decoders, where the first decoder models response generation and the second serves as a regularizer and jointly models response generation and consistency understanding. The proposed framework can benefit from the stacked encoder and decoders to learn from much smaller personalized dialogue data while maintaining competitive performance. Under different low-resource settings, subjective and objective evaluations prove that the stack-propagation framework outperforms strong baselines in response quality and persona consistency and largely overcomes the shortcomings of traditional models that rely heavily on the persona-dense dialogue data. [ABSTRACT FROM AUTHOR]

Temporal features of text have been shown to improve clustering and organization of documents, text classification, visualization, and ranking. Temporal ranking models consider the temporal expressions found in text (e.g., "in 2021" or "last year") as time units, rather than as keywords, to define a temporal relevance and improve ranking. This article introduces a new class of ranking models called Temporal Metric Space Models (TMSM), based on a new domain for representing temporal information found in documents and queries, where each temporal expression is represented as a time interval. Furthermore, we introduce a new frequency-based baseline called Temporal BM25 (TBM25). We evaluate the effectiveness of each proposed metric against a purely textual baseline, as well as several variations of the metrics themselves, where we change the aggregate function, the time granularity and the combination weight. Our extensive experiments on five test collections show statistically significant improvements of TMSM and TBM25 over state-of-the-art temporal ranking models. Combining the temporal similarity scores with the text similarity scores always improves the results, when the combination weight is between 2% and 6% for the temporal scores. This is true also for test collections where only 5% of queries contain explicit temporal expressions. [ABSTRACT FROM AUTHOR]

As recommender systems become increasingly important in daily human decision-making, users are demanding convincing explanations to understand why they get the specific recommendation results. Although a number of explainable recommender systems have recently been proposed, there still lacks an understanding of what users really need in a recommendation explanation. The actual reason behind users' intention to examine and consume (e.g., click and watch a movie) can be the window to answer this question and is named as self-explanation in this work. In addition, humans usually make recommendations accompanied by explanations, but there remain fewer studies on how humans explain and what we can learn from humangenerated explanations. To investigate these questions, we conduct a novel multi-role, multi-session user study inwhich users interact with multiple types of system-generated explanations as well as human-generated explanations, namely peer-explanation. During the study, users' intentions, expectations, and experiences are tracked in several phases, including before and after the users are presented with an explanation and after the content is examined. Through comprehensive investigations, three main findings have been made: First, we observe not only the positive but also the negative effects of explanations, and the impact varies across different types of explanations. Moreover, human-generated explanation, peer-explanation, performs better in increasing user intentions and helping users to better construct preferences, which results in better user satisfaction. Second, based on users' self-explanation, the information accuracy is measured and found to be a major factor associated with user satisfaction. Some other factors, such as unfamiliarity and similarity, are also discovered and summarized. Third, through annotations of the information aspects used in the human-generated selfexplanation and peer-explanation, patterns of how humans explain are investigated, including what information and how much information is utilized. In addition, based on the findings, a human-inspired explanation approach is proposed and found to increase user satisfaction, revealing the potential improvement of further incorporating more human patterns in recommendation explanations. These findings have shed light on the deeper understanding of the recommendation explanation and further research on its evaluation and generation. Furthermore, the collected data, including human-generated explanations by both the external peers and the users' selves, will be released to support future research works on explanation evaluation. [ABSTRACT FROM AUTHOR]