Search

Jina Embeddings constitutes a set of high-performance sentence embedding models adept at translating various textual inputs into numerical representations, thereby capturing the semantic essence of the text. While these models are not exclusively designed for text generation, they excel in applications such as dense retrieval and semantic textual similarity. This paper details the development of Jina Embeddings, starting with the creation of a high-quality pairwise and triplet dataset. It underlines the crucial role of data cleaning in dataset preparation, gives in-depth insights into the model training process, and concludes with a comprehensive performance evaluation using the Massive Textual Embedding Benchmark (MTEB)., 9 pages, 2 page appendix, EMNLP 2023 Industrial Track

Long sequences of text are challenging in the context of transformers, due to quadratic memory increase in the self-attention mechanism. As this issue directly affects the translation from natural language to SQL queries (as techniques usually take as input a concatenated text with the question and the database schema), we present techniques that allow long text sequences to be handled by transformers with up to 512 input tokens. We propose a training process with database schema pruning (removal of tables and columns names that are useless for the query of interest). In addition, we used a multilingual approach with the mT5-large model fine-tuned with a data-augmented Spider dataset in four languages simultaneously: English, Portuguese, Spanish, and French. Our proposed technique used the Spider dataset and increased the exact set match accuracy results from 0.718 to 0.736 in a validation dataset (Dev). Source code, evaluations, and checkpoints are available at: \underline{https://github.com/C4AI/gap-text2sql}., This preprint has not undergone peer review or any post-submission improvements or corrections. The Version of Record of this article is published in International Journal of Information Technology, and is available online at https://doi.org/10.1007/s41870-023-01342-3 . SharedIt link: https://rdcu.be/dff19

Recommendation systems play a vital role in many online platforms, with their primary objective being to satisfy and retain users. As directly optimizing user retention is challenging, multiple evaluation metrics are often employed. Existing methods generally formulate the optimization of these evaluation metrics as a multitask learning problem, but often overlook the fact that user preferences for different tasks are personalized and change over time. Identifying and tracking the evolution of user preferences can lead to better user retention. To address this issue, we introduce the concept of "user lifecycle", consisting of multiple stages characterized by users' varying preferences for different tasks. We propose a novel Stage-Adaptive Network (STAN) framework for modeling user lifecycle stages. STAN first identifies latent user lifecycle stages based on learned user preferences, and then employs the stage representation to enhance multi-task learning performance. Our experimental results using both public and industrial datasets demonstrate that the proposed model significantly improves multi-task prediction performance compared to state-of-the-art methods, highlighting the importance of considering user lifecycle stages in recommendation systems. Furthermore, online A/B testing reveals that our model outperforms the existing model, achieving a significant improvement of 3.05% in staytime per user and 0.88% in CVR. These results indicate that our approach effectively improves the overall efficiency of the multi-task recommendation system.

Collaborative-learning based recommender systems emerged following the success of collaborative learning techniques such as Federated Learning (FL) and Gossip Learning (GL). In these systems, users participate in the training of a recommender system while keeping their history of consumed items on their devices. While these solutions seemed appealing for preserving the privacy of the participants at a first glance, recent studies have shown that collaborative learning can be vulnerable to a variety of privacy attacks. In this paper we propose a novel privacy attack called Community Detection Attack (CDA), which allows an adversary to discover the members of a community based on a set of items of her choice (e.g., discovering users interested in LGBT content). Through experiments on three real recommendation datasets and by using two state-of-the-art recommendation models, we assess the sensitivity of an FL-based recommender system as well as two flavors of Gossip Learning-based recommender systems to CDA. Results show that on all models and all datasets, the FL setting is more vulnerable to CDA than Gossip settings. We further evaluated two off-the-shelf mitigation strategies, namely differential privacy (DP) and a share less policy, which consists in sharing a subset of model parameters. Results show a better privacy-utility trade-off for the share less policy compared to DP especially in the Gossip setting.

Search engines, such as Google, have a considerable impact on society. Therefore, undesirable consequences, such as retrieving incorrect search results, pose a risk to users. Although previous research has reported the adverse outcomes of web search, little is known about how search engine users evaluate those outcomes. In this study, we show which aspects of web search are perceived as risky using a sample (N = 3,884) representative of the German Internet population. We found that many participants are often concerned with adverse consequences immediately appearing on the search engine result page. Moreover, participants' experiences with adverse consequences are directly related to their risk perception. Our results demonstrate that people perceive risks related to web search. In addition to our study, there is a need for more independent research on the possible detrimental outcomes of web search to monitor and mitigate risks. Apart from risks for individuals, search engines with a massive number of users have an extraordinary impact on society; therefore, the acceptable risks of web search should be discussed.

Query performance prediction (QPP) is a core task in information retrieval. The QPP task is to predict the retrieval quality of a search system for a query without relevance judgments. Research has shown the effectiveness and usefulness of QPP for ad-hoc search. Recent years have witnessed considerable progress in conversational search (CS). Effective QPP could help a CS system to decide an appropriate action to be taken at the next turn. Despite its potential, QPP for CS has been little studied. We address this research gap by reproducing and studying the effectiveness of existing QPP methods in the context of CS. While the task of passage retrieval remains the same in the two settings, a user query in CS depends on the conversational history, introducing novel QPP challenges. In particular, we seek to explore to what extent findings from QPP methods for ad-hoc search generalize to three CS settings: (i) estimating the retrieval quality of different query rewriting-based retrieval methods, (ii) estimating the retrieval quality of a conversational dense retrieval method, and (iii) estimating the retrieval quality for top ranks vs. deeper-ranked lists. Our findings can be summarized as follows: (i) supervised QPP methods distinctly outperform unsupervised counterparts only when a large-scale training set is available; (ii) point-wise supervised QPP methods outperform their list-wise counterparts in most cases; and (iii) retrieval score-based unsupervised QPP methods show high effectiveness in assessing the conversational dense retrieval method, ConvDR., Accepted for publication at SIGIR 2023

In information retrieval systems, search parameters are optimized to ensure high effectiveness based on a set of past searches and these optimized parameters are then used as the system configuration for all subsequent queries. A better approach, however, would be to adapt the parameters to fit the query at hand. Selective query expansion is one such an approach, in which the system decides automatically whether or not to expand the query, resulting in two possible system configurations. This approach was extended recently to include many other parameters, leading to many possible system configurations where the system automatically selects the best configuration on a per-query basis. To determine the ideal configurations to use on a per-query basis in real-world systems we developed a method in which a restricted number of possible configurations is pre-selected and then used in a meta-search engine that decides the best search configuration on a per query basis. We define a risk-sensitive approach for configuration pre-selection that considers the risk-reward trade-off between the number of configurations kept, and system effectiveness. For final configuration selection, the decision is based on query feature similarities. We find that a relatively small number of configurations (20) selected by our risk-sensitive model is sufficient to increase effectiveness by about 15% according(P@10, nDCG@10) when compared to traditional grid search using a single configuration and by about 20% when compared to learning to rank documents. Our risk-sensitive approach works for both diversity- and ad hoc-oriented searches. Moreover, the similarity-based selection method outperforms the more sophisticated approaches. Thus, we demonstrate the feasibility of developing per-query information retrieval systems, which will guide future research in this direction., 30 pages, 5 figures, 8 tables; submitted to TOIS ACM journal

Webpages change over time, and web archives hold copies of historical versions of webpages. Users of web archives, such as journalists, want to find and view changes on webpages over time. However, the current search interfaces for web archives do not support this task. For the web archives that include a full-text search feature, multiple versions of the same webpage that match the search query are shown individually without enumerating changes, or are grouped together in a way that hides changes. We present a change text search engine that allows users to find changes in webpages. We describe the implementation of the search engine backend and frontend, including a tool that allows users to view the changes between two webpage versions in context as an animation. We evaluate the search engine with U.S. federal environmental webpages that changed between 2016 and 2020. The change text search results page can clearly show when terms and phrases were added or removed from webpages. The inverted index can also be queried to identify salient and frequently deleted terms in a corpus., In Proceedings of JCDL 2023; 20 pages, 11 figures, 2 tables

This research aims to explore various methods for assessing user feedback in mixed-initiative conversational search (CS) systems. While CS systems enjoy profuse advancements across multiple aspects, recent research fails to successfully incorporate feedback from the users. One of the main reasons for that is the lack of system-user conversational interaction data. To this end, we propose a user simulator-based framework for multi-turn interactions with a variety of mixed-initiative CS systems. Specifically, we develop a user simulator, dubbed ConvSim, that, once initialized with an information need description, is capable of providing feedback to a system's responses, as well as answering potential clarifying questions. Our experiments on a wide variety of state-of-the-art passage retrieval and neural re-ranking models show that effective utilization of user feedback can lead to 16% retrieval performance increase in terms of nDCG@3. Moreover, we observe consistent improvements as the number of feedback rounds increases (35% relative improvement in terms of nDCG@3 after three rounds). This points to a research gap in the development of specific feedback processing modules and opens a potential for significant advancements in CS. To support further research in the topic, we release over 30,000 transcripts of system-simulator interactions based on well-established CS datasets., 11 pages, 2 figures, to be published in SIGIR 2023

Current metric learning approaches for image retrieval are usually based on learning a space of informative latent representations where simple approaches such as the cosine distance will work well. Recent state of the art methods such as HypViT move to more complex embedding spaces that may yield better results but are harder to scale to production environments. In this work, we first construct a simpler model based on triplet loss with hard negatives mining that performs at the state of the art level but does not have these drawbacks. Second, we introduce a novel approach for image retrieval postprocessing called Siamese Transformer for Image Retrieval (STIR) that reranks several top outputs in a single forward pass. Unlike previously proposed Reranking Transformers, STIR does not rely on global/local feature extraction and directly compares a query image and a retrieved candidate on pixel level with the usage of attention mechanism. The resulting approach defines a new state of the art on standard image retrieval datasets: Stanford Online Products and DeepFashion In-shop. We also release the source code at https://github.com/OML-Team/open-metric-learning/tree/main/pipelines/postprocessing/ and an interactive demo of our approach at https://dapladoc-oml-postprocessing-demo-srcappmain-pfh2g0.streamlit.app/, 14 pages, 3 figures

Evaluating the performance of software for automated vehicles is predominantly driven by data collected from the real world. While professional test drivers are supported with technical means to semi-automatically annotate driving maneuvers to allow better event identification, simple data loggers in large vehicle fleets typically lack automatic and detailed event classification and hence, extra effort is needed when post-processing such data. Yet, the data quality from professional test drivers is apparently higher than the one from large fleets where labels are missing, but the non-annotated data set from large vehicle fleets is much more representative for typical, realistic driving scenarios to be handled by automated vehicles. However, while growing the data from large fleets is relatively simple, adding valuable annotations during post-processing has become increasingly expensive. In this paper, we leverage Z-order space-filling curves to systematically reduce data dimensionality while preserving domain-specific data properties, which allows us to explore even large-scale field data sets to spot interesting events orders of magnitude faster than processing time-series data directly. Furthermore, the proposed concept is based on an analytical approach, which preserves explainability for the identified events.

When asked, current large language models (LLMs) like ChatGPT claim that they can assist us with relevance judgments. Many researchers think this would not lead to credible IR research. In this perspective paper, we discuss possible ways for LLMs to assist human experts along with concerns and issues that arise. We devise a human-machine collaboration spectrum that allows categorizing different relevance judgment strategies, based on how much the human relies on the machine. For the extreme point of "fully automated assessment", we further include a pilot experiment on whether LLM-based relevance judgments correlate with judgments from trained human assessors. We conclude the paper by providing two opposing perspectives - for and against the use of LLMs for automatic relevance judgments - and a compromise perspective, informed by our analyses of the literature, our preliminary experimental evidence, and our experience as IR researchers. We hope to start a constructive discussion within the community to avoid a stale-mate during review, where work is dammed if is uses LLMs for evaluation and dammed if it doesn't.

Some recent \textit{news recommendation} (NR) methods introduce a Pre-trained Language Model (PLM) to encode news representation by following the vanilla pre-train and fine-tune paradigm with carefully-designed recommendation-specific neural networks and objective functions. Due to the inconsistent task objective with that of PLM, we argue that their modeling paradigm has not well exploited the abundant semantic information and linguistic knowledge embedded in the pre-training process. Recently, the pre-train, prompt, and predict paradigm, called \textit{prompt learning}, has achieved many successes in natural language processing domain. In this paper, we make the first trial of this new paradigm to develop a \textit{Prompt Learning for News Recommendation} (Prompt4NR) framework, which transforms the task of predicting whether a user would click a candidate news as a cloze-style mask-prediction task. Specifically, we design a series of prompt templates, including discrete, continuous, and hybrid templates, and construct their corresponding answer spaces to examine the proposed Prompt4NR framework. Furthermore, we use the prompt ensembling to integrate predictions from multiple prompt templates. Extensive experiments on the MIND dataset validate the effectiveness of our Prompt4NR with a set of new benchmark results., 11 pages, 7 figures, 3 tables, SIGIR2023 conference

The paper describes a transformer-based system designed for SemEval-2023 Task 9: Multilingual Tweet Intimacy Analysis. The purpose of the task was to predict the intimacy of tweets in a range from 1 (not intimate at all) to 5 (very intimate). The official training set for the competition consisted of tweets in six languages (English, Spanish, Italian, Portuguese, French, and Chinese). The test set included the given six languages as well as external data with four languages not presented in the training set (Hindi, Arabic, Dutch, and Korean). We presented a solution based on an ensemble of XLM-T, a multilingual RoBERTa model adapted to the Twitter domain. To improve the performance of unseen languages, each tweet was supplemented by its English translation. We explored the effectiveness of translated data for the languages seen in fine-tuning compared to unseen languages and estimated strategies for using translated data in transformer-based models. Our solution ranked 4th on the leaderboard while achieving an overall Pearson's r of 0.599 over the test set. The proposed system improves up to 0.088 Pearson's r over a score averaged across all 45 submissions., 7 pages. The 17th International Workshop on Semantic Evaluation (SemEval-2023)

The advent of personalized news recommendation has given rise to increasingly complex recommender architectures. Most neural news recommenders rely on user click behavior and typically introduce dedicated user encoders that aggregate the content of clicked news into user embeddings (early fusion). These models are predominantly trained with standard point-wise classification objectives. The existing body of work exhibits two main shortcomings: (1) despite general design homogeneity, direct comparisons between models are hindered by varying evaluation datasets and protocols; (2) it leaves alternative model designs and training objectives vastly unexplored. In this work, we present a unified framework for news recommendation, allowing for a systematic and fair comparison of news recommenders across several crucial design dimensions: (i) candidate-awareness in user modeling, (ii) click behavior fusion, and (iii) training objectives. Our findings challenge the status quo in neural news recommendation. We show that replacing sizable user encoders with parameter-efficient dot products between candidate and clicked news embeddings (late fusion) often yields substantial performance gains. Moreover, our results render contrastive training a viable alternative to point-wise classification objectives., Accepted at the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2023)

Nowadays, false and unverified information on social media sway individuals' perceptions during major geo-political events and threaten the quality of the whole digital information ecosystem. Since the Russian invasion of Ukraine, several fact-checking organizations have been actively involved in verifying stories related to the conflict that circulated online. In this paper, we leverage a public repository of fact-checked claims to build a methodological framework for automatically identifying false and unsubstantiated claims spreading on Twitter in February 2022. Our framework consists of two sequential models: First, the claim detection model identifies whether tweets incorporate a (false) claim among those considered in our collection. Then, the claim retrieval model matches the tweets with fact-checked information by ranking verified claims according to their relevance with the input tweet. Both models are based on pre-trained language models and fine-tuned to perform a text classification task and an information retrieval task, respectively. In particular, to validate the effectiveness of our methodology, we consider 83 verified false claims that spread on Twitter during the first week of the invasion, and manually annotate 5,872 tweets according to the claim(s) they report. Our experiments show that our proposed methodology outperforms standard baselines for both claim detection and claim retrieval. Overall, our results highlight how social media providers could effectively leverage semi-automated approaches to identify, track, and eventually moderate false information that spreads on their platforms., 7 pages, 2 figures, WWW23 Companion Proceedings

Approximate K nearest neighbor (AKNN) search is a fundamental and challenging problem. We observe that in high-dimensional space, the time consumption of nearly all AKNN algorithms is dominated by that of the distance comparison operations (DCOs). For each operation, it scans full dimensions of an object and thus, runs in linear time wrt the dimensionality. To speed it up, we propose a randomized algorithm named ADSampling which runs in logarithmic time wrt to the dimensionality for the majority of DCOs and succeeds with high probability. In addition, based on ADSampling we develop one general and two algorithm-specific techniques as plugins to enhance existing AKNN algorithms. Both theoretical and empirical studies confirm that: (1) our techniques introduce nearly no accuracy loss and (2) they consistently improve the efficiency., This paper has been accepted by SIGMOD 2023

Interleaving is an online evaluation approach for information retrieval systems that compares the effectiveness of ranking functions in interpreting the users' implicit feedback. Previous work such as Hofmann et al (2011) has evaluated the most promising interleaved methods at the time, on uniform distributions of queries. In the real world, ordinarily, there is an unbalanced distribution of repeated queries that follows a long-tailed users' search demand curve. The more a query is executed, by different users (or in different sessions), the higher the probability of collecting implicit feedback (interactions/clicks) on the related search results. This paper first aims to replicate the Team Draft Interleaving accuracy evaluation on uniform query distributions and then focuses on assessing how this method generalizes to long-tailed real-world scenarios. The reproducibility work raised interesting considerations on how the winning ranking function for each query should impact the overall winner for the entire evaluation. Based on what was observed, we propose that not all the queries should contribute to the final decision in equal proportion. As a result of these insights, we designed two variations of the $\Delta_{AB}$ score winner estimator that assign to each query a credit based on statistical hypothesis testing. To replicate, reproduce and extend the original work, we have developed from scratch a system that simulates a search engine and users' interactions from datasets from the industry. Our experiments confirm our intuition and show that our methods are promising in terms of accuracy, sensitivity, and robustness to noise., Comment: This preprint has not undergone peer review (when applicable) or any post-submission improvements or corrections. The Version of Record of this contribution is published in Advances in Information Retrieval 45th European Conference on Information Retrieval, ECIR 2023, Dublin, Ireland, April, 2023, Proceedings, Part III, and is available online at https://doi.org/10.1007/978-3-031-28241-6_2

In many applications of X-ray computed tomography, an unsupervised segmentation of the reconstructed 3D volumes forms an important step in the image processing chain for further investigation of the digitized object. Therefore, the goal is to train a clustering algorithm on the volume, which produces a voxelwise classification by assigning a cluster index to each voxel. However, clustering methods, e.g., K-Means, typically have an asymptotic polynomial runtime with respect to the dataset size, and thus, these techniques are rarely applicable to large volumes. In this work, we introduce a novel clustering technique based on random sampling, which allows for the voxelwise classification of arbitrarily large volumes. The presented method conducts efficient linear passes over the data to extract a representative random sample of a fixed size on which the classifier can be trained. Then, a final linear pass performs the segmentation and assigns a cluster index to each individual voxel. Quantitative and qualitative evaluations show that excellent results can be achieved even with a very small sample size. Consequently, the unsupervised segmentation by means of clustering becomes feasible for arbitrarily large volumes., 12 pages, 8 figures

Scientific press briefings are a valuable information source. They consist of alternating expert speeches, questions from the audience and their answers. Therefore, they can contribute to scientific and fact-based media coverage. Even though press briefings are highly informative, extracting statements relevant to individual journalistic tasks is challenging and time-consuming. To support this task, an automated statement extraction system is proposed. Claims are used as the main feature to identify statements in press briefing transcripts. The statement extraction task is formulated as a four-step procedure. First, the press briefings are split into sentences and passages, then claim sentences are identified through sequence classification. Subsequently, topics are detected, and the sentences are filtered to improve the coherence and assess the length of the statements. The results indicate that claim detection can be used to identify statements in press briefings. While many statements can be extracted automatically with this system, they are not always as coherent as needed to be understood without context and may need further review by knowledgeable persons., Datenbanksysteme f\"ur Business, Technologie und Web (BTW 2023)

People use search engines to find answers to questions related to their health, finances, or other socially relevant issues. However, most users are unaware that search results are considerably influenced by search engine marketing (SEM). SEM measures are driven by commercial, political, or other motives. Due to these motivations, two questions arise: What information quality is mediated through SEM? And how is collecting documents of different quality affecting user knowledge gain? Both questions are not considered by existing models of information behavior. Hence, the doctoral research project described in this paper aims to develop and empirically test an information search behavior model on the influences of SEM on user knowledge gain and thereby contribute to the search as learning body of research.

Language models have become increasingly popular in recent years for tasks like information retrieval. As use-cases become oriented toward specific domains, fine-tuning becomes default for standard performance. To fine-tune these models for specific tasks and datasets, it is necessary to carefully tune the model's hyperparameters and training techniques. In this paper, we present an in-depth analysis of the performance of four transformer-based language models on the task of biomedical information retrieval. The models we consider are DeepMind's RETRO (7B parameters), GPT-J (6B parameters), GPT-3 (175B parameters), and BLOOM (176B parameters). We compare their performance on the basis of relevance, accuracy, and interpretability, using a large corpus of 480000 research papers on protein structure/function prediction as our dataset. Our findings suggest that smaller models, with, 9 pages, 4 figures

Ranking systems are ubiquitous in modern Internet services, including online marketplaces, social media, and search engines. Traditionally, ranking systems only focus on how to get better relevance estimation. When relevance estimation is available, they usually adopt a user-centric optimization strategy where ranked lists are generated by sorting items according to their estimated relevance. However, such user-centric optimization ignores the fact that item providers also draw utility from ranking systems. It has been shown in existing research that such user-centric optimization will cause much unfairness to item providers, followed by unfair opportunities and unfair economic gains for item providers. To address ranking fairness, many fair ranking methods have been proposed. However, as we show in this paper, these methods could be suboptimal as they directly rely on the relevance estimation without being aware of the uncertainty (i.e., the variance of the estimated relevance). To address this uncertainty, we propose a novel Marginal-Certainty-aware Fair algorithm named MCFair. MCFair jointly optimizes fairness and user utility, while relevance estimation is constantly updated in an online manner. In MCFair, we first develop a ranking objective that includes uncertainty, fairness, and user utility. Then we directly use the gradient of the ranking objective as the ranking score. We theoretically prove that MCFair based on gradients is optimal for the aforementioned ranking objective. Empirically, we find that on semi-synthesized datasets, MCFair is effective and practical and can deliver superior performance compared to state-of-the-art fair ranking methods. To facilitate reproducibility, we release our code https://github.com/Taosheng-ty/WSDM22-MCFair., 10 pages, 5 figures

Metrics for merge trees that are simultaneously stable, informative, and efficiently computable have so far eluded researchers. We show in this work that it is possible to devise such a metric when restricting merge trees to ordered domains such as the interval and the circle. We present the ``dynamic ordered persistence editing'' (DOPE) distance, which we prove is stable and informative while satisfying metric properties. We then devise a simple $O(N^2)$ dynamic programming algorithm to compute it on the interval and an $O(N^3)$ algorithm to compute it on the circle. Surprisingly, we accomplish this by ignoring all of the hierarchical information of the merge tree and simply focusing on a sequence of ordered critical points, which can be interpreted as a time series. Thus our algorithm is more similar to string edit distance and dynamic time warping than it is to more conventional merge tree comparison algorithms. In the context of time series with the interval as a domain, we show empirically on the UCR time series classification dataset that DOPE performs better than bottleneck/Wasserstein distances between persistence diagrams., 31 pages, 12 Figures

An understanding of how Social Overhead Capital (SOC) is associated with the land value of the local community is important for effective urban planning. However, even within a district, there are multiple sections used for different purposes; the term for this is spatial heterogeneity. The spatial heterogeneity issue has to be considered when attempting to comprehend land prices. If there is spatial heterogeneity within a district, land prices can be managed by adopting the spatial clustering method. In this study, spatial attributes including SOC, socio-demographic features, and spatial information in a specific district are analyzed with Finite Mixture Modeling (FMM) in order to find (a) the optimal number of clusters and (b) the association among SOCs, socio-demographic features, and land prices. FMM is a tool used to find clusters and the attributes' coefficients simultaneously. Using the FMM method, the results show that four clusters exist in one district and the four clusters have different associations among SOCs, demographic features, and land prices. Policymakers and managerial administration need to look for information to make policy about land prices. The current study finds the consideration of closeness to SOC to be a significant factor on land prices and suggests the potential policy direction related to SOC., 26 pages, 3 figures

Extracting useful information from the user history to clearly understand informational needs is a crucial feature of a proactive information retrieval system. Regarding understanding information and relevance, Wikipedia can provide the background knowledge that an intelligent system needs. This work explores how exploiting the context of a query using Wikipedia concepts can improve proactive information retrieval on noisy text. We formulate two models that use entity linking to associate Wikipedia topics with the relevance model. Our experiments around a podcast segment retrieval task demonstrate that there is a clear signal of relevance in Wikipedia concepts while a ranking model can improve precision by incorporating them. We also find Wikifying the background context of a query can help disambiguate the meaning of the query, further helping proactive information retrieval., To be published at the First Workshop on Proactive and Agent-Supported Information Retrieval at CIKM 2022

Question generation has recently gained a lot of research interest, especially with the advent of large language models. In and of itself, question generation can be considered 'AI-hard', as there is a lack of unanimously agreed sense of what makes a question 'good' or 'bad'. In this paper, we tackle two fundamental problems in parallel: on one hand, we try to solve the scaling problem, where question-generation and answering applications have to be applied to a massive amount of text without ground truth labeling. The usual approach to solve this problem is to either downsample or summarize. However, there are critical risks of misinformation with these approaches. On the other hand, and related to the misinformation problem, we try to solve the 'safety' problem, as many public institutions rely on a much higher level of accuracy for the content they provide. We introduce an adversarial approach to tackle the question generation safety problem with scale. Specifically, we designed a question-answering system that specifically prunes out unanswerable questions that may be generated, and further increases the quality of the answers that are generated. We build a production-ready, easily-plugged pipeline that can be used on any given body of text, that is scalable and immune from generating any hate speech, profanity, or misinformation. Based on the results, we are able to generate more than six times the number of quality questions generated by the abstractive approach, with a perceived quality being 44% higher, according to a survey of 168 participants., 15 pages, 8 figures, 2 tables

The search for research datasets is as important as laborious. Due to the importance of the choice of research data in further research, this decision must be made carefully. Additionally, because of the growing amounts of data in almost all areas, research data is already a central artifact in empirical sciences. Consequentially, research dataset recommendations can beneficially supplement scientific publication searches. We formulated the recommendation task as a retrieval problem by focussing on broad similarities between research datasets and scientific publications. In a multistage approach, initial recommendations were retrieved by the BM25 ranking function and dynamic queries. Subsequently, the initial ranking was re-ranked utilizing click feedback and document embeddings. The proposed system was evaluated live on real user interaction data using the STELLA infrastructure in the LiLAS Lab at CLEF 2021. Our experimental system could efficiently be fine-tuned before the live evaluation by pre-testing the system with a pseudo test collection based on prior user interaction data from the live system. The results indicate that the experimental system outperforms the other participating systems., Best of 2021 Labs: LiLAS

Retrieval-augmented models have proven to be effective in natural language processing tasks, yet there remains a lack of research on their optimization using variational inference. We introduce the Variational Open-Domain (VOD) framework for end-to-end training and evaluation of retrieval-augmented models, focusing on open-domain question answering and language modelling. The VOD objective, a self-normalized estimate of the R\'enyi variational bound, approximates the task marginal likelihood and is evaluated under samples drawn from an auxiliary sampling distribution (cached retriever and/or approximate posterior). It remains tractable, even for retriever distributions defined on large corpora. We demonstrate VOD's versatility by training reader-retriever BERT-sized models on multiple-choice medical exam questions. On the MedMCQA dataset, we outperform the domain-tuned Med-PaLM by +5.3% despite using 2.500$\times$ fewer parameters. Our retrieval-augmented BioLinkBERT model scored 62.9% on the MedMCQA and 55.0% on the MedQA-USMLE. Last, we show the effectiveness of our learned retriever component in the context of medical semantic search., Comment: 28 pages, 5 figures. Accepted at ICML 2023

The paper describes neural models developed for the DAGPap22 shared task hosted at the Third Workshop on Scholarly Document Processing. This shared task targets the automatic detection of generated scientific papers. Our work focuses on comparing different transformer-based models as well as using additional datasets and techniques to deal with imbalanced classes. As a final submission, we utilized an ensemble of SciBERT, RoBERTa, and DeBERTa fine-tuned using random oversampling technique. Our model achieved 99.24% in terms of F1-score. The official evaluation results have put our system at the third place., Accepted to SDP 2022 (Third Workshop on Scholarly Document Processing collocated with COLING 2022)

BERT-style models pre-trained on the general corpus (e.g., Wikipedia) and fine-tuned on specific task corpus, have recently emerged as breakthrough techniques in many NLP tasks: question answering, text classification, sequence labeling and so on. However, this technique may not always work, especially for two scenarios: a corpus that contains very different text from the general corpus Wikipedia, or a task that learns embedding spacial distribution for a specific purpose (e.g., approximate nearest neighbor search). In this paper, to tackle the above two scenarios that we have encountered in an industrial e-commerce search system, we propose customized and novel pre-training tasks for two critical modules: user intent detection and semantic embedding retrieval. The customized pre-trained models after fine-tuning, being less than 10% of BERT-base's size in order to be feasible for cost-efficient CPU serving, significantly improve the other baseline models: 1) no pre-training model and 2) fine-tuned model from the official pre-trained BERT using general corpus, on both offline datasets and online system. We have open sourced our datasets for the sake of reproducibility and future works., 5 pages, 3 figures; accepted by CIKM2022

A sequence of recent papers has considered the role of measurement scales in information retrieval (IR) experimentation, and presented the argument that (only) uniform-step interval scales should be used, and hence that well-known metrics such as reciprocal rank, expected reciprocal rank, normalized discounted cumulative gain, and average precision, should be either discarded as measurement tools, or adapted so that their metric values lie at uniformly-spaced points on the number line. These papers paint a rather bleak picture of past decades of IR evaluation, at odds with the community's overall emphasis on practical experimentation and measurable improvement. Our purpose in this work is to challenge that position. In particular, we argue that mappings from categorical and ordinal data to sets of points on the number line are valid provided there is an external reason for each target point to have been selected. We first consider the general role of measurement scales, and of categorical, ordinal, interval, ratio, and absolute data collections. In connection with the first two of those categories we also provide examples of the knowledge that is captured and represented by numeric mappings to the real number line. Focusing then on information retrieval, we argue that document rankings are categorical data, and that the role of an effectiveness metric is to provide a single value that represents the usefulness to a user or population of users of any given ranking, with usefulness able to be represented as a continuous variable on a ratio scale. That is, we argue that current IR metrics are well-founded, and, moreover, that those metrics are more meaningful in their current form than in the proposed "intervalized" versions.

Timeliness and contextual accuracy of recommendations are increasingly important when delivering contemporary digital marketing experiences. Conventional recommender systems (RS) suggest relevant but time-invariant items to users by accounting for their past purchases. These recommendations only map to customers' general preferences rather than a customer's specific needs immediately preceding a purchase. In contrast, RSs that consider the order of transactions, purchases, or experiences to measure evolving preferences can offer more salient and effective recommendations to customers: Sequential RSs not only benefit from a better behavioral understanding of a user's current needs but also better predictive power. In this paper, we demonstrate and rank the effectiveness of a sequential recommendation system by utilizing a production dataset of over 2.7 million credit card transactions for 46K cardholders. The method first employs an autoencoder on raw transaction data and submits observed transaction encodings to a GRU-based sequential model. The sequential model produces a MAP@1 metric of 47% on the out-of-sample test set, in line with existing research. We also discuss implications for embedding real-time predictions using the sequential RS into Nexus, a scalable, low-latency, event-based digital experience architecture., 18 pages, 9 figures

Session-based recommendation (SBR) aims to predict the user next action based on the ongoing sessions. Recently, there has been an increasing interest in modeling the user preference evolution to capture the fine-grained user interests. While latent user preferences behind the sessions drift continuously over time, most existing approaches still model the temporal session data in discrete state spaces, which are incapable of capturing the fine-grained preference evolution and result in sub-optimal solutions. To this end, we propose Graph Nested GRU ordinary differential equation (ODE), namely GNG-ODE, a novel continuum model that extends the idea of neural ODEs to continuous-time temporal session graphs. The proposed model preserves the continuous nature of dynamic user preferences, encoding both temporal and structural patterns of item transitions into continuous-time dynamic embeddings. As the existing ODE solvers do not consider graph structure change and thus cannot be directly applied to the dynamic graph, we propose a time alignment technique, called t-Alignment, to align the updating time steps of the temporal session graphs within a batch. Empirical results on three benchmark datasets show that GNG-ODE significantly outperforms other baselines., Under Review

Animated avatars, which look and talk like humans, are iconic visions of the future of AI-powered systems. Through many sci-fi movies, we are acquainted with the idea of speaking to such virtual personalities as if they were humans. Today, we talk more and more to machines like Apple's Siri, for example, to ask them for the weather forecast. However, when asked for recommendations, for example, for a restaurant to go to, the limitations of such devices quickly become obvious. They do not engage in a conversation to find out what we might prefer, they often do not provide explanations for what they recommend, and they may have difficulties remembering what was said 1 min earlier. Conversational recommender systems (CRS) promise to address these limitations. In this paper, we review existing approaches to building such systems, which developments we observe today, which challenges are still open and why the development of conversational recommenders represents one of the next grand challenges of AI.

This work explores how population-based engagement prediction can address cold-start at scale in large learning resource collections. The paper introduces i) VLE, a novel dataset that consists of content and video based features extracted from publicly available scientific video lectures coupled with implicit and explicit signals related to learner engagement, ii) two standard tasks related to predicting and ranking context-agnostic engagement in video lectures with preliminary baselines and iii) a set of experiments that validate the usefulness of the proposed dataset. Our experimental results indicate that the newly proposed VLE dataset leads to building context-agnostic engagement prediction models that are significantly performant than ones based on previous datasets, mainly attributing to the increase of training examples. VLE dataset's suitability in building models towards Computer Science/ Artificial Intelligence education focused on e-learning/ MOOC use-cases is also evidenced. Further experiments in combining the built model with a personalising algorithm show promising improvements in addressing the cold-start problem encountered in educational recommenders. This is the largest and most diverse publicly available dataset to our knowledge that deals with learner engagement prediction tasks. The dataset, helper tools, descriptive statistics and example code snippets are available publicly., To be presented at International Conference for Educational Data Mining 2022

This report provides an overview of the field of Information Retrieval (IR) in healthcare. It does not aim to introduce general concepts and theories of IR but to present and describe specific aspects of Health Information Retrieval (HIR). After a brief introduction to the more broader field of IR, the significance of HIR at current times is discussed. Specific characteristics of Health Information, its classification and the main existing representations for health concepts are described together with the main products and services in the area (e.g.: databases of health bibliographic content, health specific search engines and others). Recent research work is discussed and the most active researchers, projects and research groups are also presented. Main organizations and journals are also identified., 38 pages, 0 figures

In this paper, we present a cross-modal recipe retrieval framework, Transformer-based Network for Large Batch Training (TNLBT), which is inspired by ACME~(Adversarial Cross-Modal Embedding) and H-T~(Hierarchical Transformer). TNLBT aims to accomplish retrieval tasks while generating images from recipe embeddings. We apply the Hierarchical Transformer-based recipe text encoder, the Vision Transformer~(ViT)-based recipe image encoder, and an adversarial network architecture to enable better cross-modal embedding learning for recipe texts and images. In addition, we use self-supervised learning to exploit the rich information in the recipe texts having no corresponding images. Since contrastive learning could benefit from a larger batch size according to the recent literature on self-supervised learning, we adopt a large batch size during training and have validated its effectiveness. In the experiments, the proposed framework significantly outperformed the current state-of-the-art frameworks in both cross-modal recipe retrieval and image generation tasks on the benchmark Recipe1M. This is the first work which confirmed the effectiveness of large batch training on cross-modal recipe embeddings., Accepted at MMM2023

CODEC is a document and entity ranking benchmark that focuses on complex research topics. We target essay-style information needs of social science researchers, i.e. "How has the UK's Open Banking Regulation benefited Challenger Banks?". CODEC includes 42 topics developed by researchers and a new focused web corpus with semantic annotations including entity links. This resource includes expert judgments on 17,509 documents and entities (416.9 per topic) from diverse automatic and interactive manual runs. The manual runs include 387 query reformulations, providing data for query performance prediction and automatic rewriting evaluation. CODEC includes analysis of state-of-the-art systems, including dense retrieval and neural re-ranking. The results show the topics are challenging with headroom for document and entity ranking improvement. Query expansion with entity information shows significant gains in document ranking, demonstrating the resource's value for evaluating and improving entity-oriented search. We also show that the manual query reformulations significantly improve document ranking and entity ranking performance. Overall, CODEC provides challenging research topics to support the development and evaluation of entity-centric search methods., 10 pages, SIGIR 2022 Preprint

One of the first steps in many text-based social science studies is to retrieve documents that are relevant for the analysis from large corpora of otherwise irrelevant documents. The conventional approach in social science to address this retrieval task is to apply a set of keywords and to consider those documents to be relevant that contain at least one of the keywords. But the application of incomplete keyword lists risks drawing biased inferences. More complex and costly methods such as query expansion techniques, topic model-based classification rules, and active as well as passive supervised learning could have the potential to more accurately separate relevant from irrelevant documents and thereby reduce the potential size of bias. Yet, whether applying these more expensive approaches increases retrieval performance compared to keyword lists at all, and if so, by how much, is unclear as a comparison of these approaches is lacking. This study closes this gap by comparing these methods across three retrieval tasks associated with a data set of German tweets (Linder, 2017), the Social Bias Inference Corpus (SBIC) (Sap et al., 2020), and the Reuters-21578 corpus (Lewis, 1997). Results show that query expansion techniques and topic model-based classification rules in most studied settings tend to decrease rather than increase retrieval performance. Active supervised learning, however, if applied on a not too small set of labeled training instances (e.g. 1,000 documents), reaches a substantially higher retrieval performance than keyword lists., 78 pages, 17 figures, 9 tables

We propose a simple model of an idealized online cultural market in which $N$ items, endowed with a hidden quality metric, are recommended to users by a ranking algorithm possibly biased by the current items' popularity. Our goal is to better understand the underlying mechanisms of the well-known fact that popularity bias can prevent higher-quality items from becoming more popular than lower-quality items, producing an undesirable misalignment between quality and popularity rankings. We do so under the assumption that users, having limited time/attention, are able to discriminate the best-quality only within a random subset of the items. We discover the existence of a harmful regime in which improper use of popularity can seriously compromise the emergence of quality, and a benign regime in which wise use of popularity, coupled with a small discrimination effort on behalf of users, guarantees the perfect alignment of quality and popularity ranking. Our findings clarify the effects of algorithmic popularity bias on quality outcomes, and may inform the design of more principled mechanisms for techno-social cultural markets., ACM Trans. Inf. Syst. April 2022

Recommendation is the task of ranking items (e.g. movies or products) according to individual user needs. Current systems rely on collaborative filtering and content-based techniques, which both require structured training data. We propose a framework for recommendation with off-the-shelf pretrained language models (LM) that only used unstructured text corpora as training data. If a user $u$ liked \textit{Matrix} and \textit{Inception}, we construct a textual prompt, e.g. \textit{"Movies like Matrix, Inception, ${}$"} to estimate the affinity between $u$ and $m$ with LM likelihood. We motivate our idea with a corpus analysis, evaluate several prompt structures, and we compare LM-based recommendation with standard matrix factorization trained on different data regimes. The code for our experiments is publicly available (https://colab.research.google.com/drive/1f1mlZ-FGaLGdo5rPzxf3vemKllbh2esT?usp=sharing)., Comment: Accepted at ECIR 2022

Dense retrieval has shown promising results in many information retrieval (IR) related tasks, whose foundation is high-quality text representation learning for effective search. Some recent studies have shown that autoencoder-based language models are able to boost the dense retrieval performance using a weak decoder. However, we argue that 1) it is not discriminative to decode all the input texts and, 2) even a weak decoder has the bypass effect on the encoder. Therefore, in this work, we introduce a novel contrastive span prediction task to pre-train the encoder alone, but still retain the bottleneck ability of the autoencoder. % Therefore, in this work, we propose to drop out the decoder and introduce a novel contrastive span prediction task to pre-train the encoder alone. The key idea is to force the encoder to generate the text representation close to its own random spans while far away from others using a group-wise contrastive loss. In this way, we can 1) learn discriminative text representations efficiently with the group-wise contrastive learning over spans and, 2) avoid the bypass effect of the decoder thoroughly. Comprehensive experiments over publicly available retrieval benchmark datasets show that our approach can outperform existing pre-training methods for dense retrieval significantly., Accepted to SIGIR 2022

While any infimum in a poset can also be computed as a supremum, and vice versa, categorical limits and colimits do not always approximate each other. If I approach a point from below, and you approach it from above, then we will surely meet if we live in a poset, but we may miss each other in a category. Can we characterize the limits and the colimits that approximate each other, and guarantee that we will meet? Such limits and colimits are called *tight*. Some critically important network applications depend on them. This paper characterizes tight limits and colimits, and describes tight completions, derived by applying the nucleus construction to adjunctions between loose completions. Just as the Dedekind-MacNeille completion of a poset preserves any existing infima and suprema, the tight completion of a category preserves any existing tight limits and colimits and is therefore idempotent., 48 pages, 16 page appendix, lots of diagrams

Background. From information theory, surprisal is a measurement of how unexpected an event is. Statistical language models provide a probabilistic approximation of natural languages, and because surprisal is constructed with the probability of an event occuring, it is therefore possible to determine the surprisal associated with English sentences. The issues and pull requests of software repository issue trackers give insight into the development process and likely contain the surprising events of this process. Objective. Prior works have identified that unusual events in software repositories are of interest to developers, and use simple code metrics-based methods for detecting them. In this study we will propose a new method for unusual event detection in software repositories using surprisal. With the ability to find surprising issues and pull requests, we intend to further analyse them to determine if they actually hold importance in a repository, or if they pose a significant challenge to address. If it is possible to find bad surprises early, or before they cause additional troubles, it is plausible that effort, cost and time will be saved as a result. Method. After extracting the issues and pull requests from 5000 of the most popular software repositories on GitHub, we will train a language model to represent these issues. We will measure their perceived importance in the repository, measure their resolution difficulty using several analogues, measure the surprisal of each, and finally generate inferential statistics to describe any correlations., 8 pages, 1 figure. Submitted to 2022 International Conference on Mining Software Repositories Registered Reports track

Within the context of review analytics, aspects are the features of products and services at which customers target their opinions and sentiments. Aspect detection helps product owners and service providers to identify shortcomings and prioritize customers' needs, and hence, maintain revenues and mitigate customer churn. Existing methods focus on detecting the surface form of an aspect by training supervised learning methods that fall short when aspects are latent in reviews. In this paper, we propose an unsupervised method to extract latent occurrences of aspects. Specifically, we assume that a customer undergoes a two-stage hypothetical generative process when writing a review: (1) deciding on an aspect amongst the set of aspects available for the product or service, and (2) writing the opinion words that are more interrelated to the chosen aspect from the set of all words available in a language. We employ latent Dirichlet allocation to learn the latent aspects distributions for generating the reviews. Experimental results on benchmark datasets show that our proposed method is able to improve the state of the art when the aspects are latent with no surface form in reviews., 6 pages, 2 figures, 2 tables

Food production is a complex process which can benefit from many optimisation approaches. However, there is growing interest in methods that support customisation of food properties to satisfy individual consumer preferences. This paper addresses the personalisation of beer properties. Having identified components of the production process for craft beers whose production tends to be less standardised, we introduce a system which enables brewers to map the desired beer properties into ingredients dosage and combination. Previously explored approaches include direct use of structural equations as well as global machine learning methods. We introduce a framework which uses an evolutionary method supporting multi-objective optimisation. This work identifies problem-dependent objectives, their associations, and proposes a workflow to automate the discovery of multiple novel recipes based on user-defined criteria. The quality of the solutions generated by the multi-objective optimiser is compared against solutions from multiple runs of the method, and those of a single objective evolutionary technique. This comparison provides a road-map allowing the users to choose among more varied options or to fine-tune one of the favourite identified solution. The experiments presented here demonstrate the usability of the framework as well as the transparency of its criteria., 13 pages, 3 figures, 6 tables, GECCO

In recent years, with the increase of social investment in scientific research, the number of research results in various fields has increased significantly. Accurately and effectively predicting the trends of future research topics can help researchers discover future research hotspots. However, due to the increasingly close correlation between various research themes, there is a certain dependency relationship between a large number of research themes. Viewing a single research theme in isolation and using traditional sequence problem processing methods cannot effectively explore the spatial dependencies between these research themes. To simultaneously capture the spatial dependencies and temporal changes between research topics, we propose a deep neural network-based research topic hotness prediction algorithm, a spatiotemporal convolutional network model. Our model combines a graph convolutional neural network (GCN) and Temporal Convolutional Network (TCN), specifically, GCNs are used to learn the spatial dependencies of research topics a and use space dependence to strengthen spatial characteristics. TCN is used to learn the dynamics of research topics' trends. Optimization is based on the calculation of weighted losses based on time distance. Compared with the current mainstream sequence prediction models and similar spatiotemporal models on the paper datasets, experiments show that, in research topic prediction tasks, our model can effectively capture spatiotemporal relationships and the predictions outperform state-of-art baselines., 11 pages,3 figures

In this brief, we conduct a complex-network analysis of the Bitcoin transaction network. In particular, we design a new sampling method, namely random walk with flying-back (RWFB), to conduct effective data sampling. We then conduct a comprehensive analysis of the Bitcoin network in terms of the degree distribution, clustering coefficient, the shortest-path length, connected component, centrality, assortativity, and the rich-club coefficient. We obtain several important observations including the small-world phenomenon, multi-center status, preferential attachment, and non-rich-club effect of the current network. This work brings up an in-depth understanding of the current Bitcoin blockchain network and offers implications for future directions in malicious activity and fraud detection in cryptocurrency blockchain networks., 6 pages, 4 figures

Word embeddings, made widely popular in 2013 with the release of word2vec, have become a mainstay of NLP engineering pipelines. Recently, with the release of BERT, word embeddings have moved from the term-based embedding space to the contextual embedding space -- each term is no longer represented by a single low-dimensional vector but instead each term and \emph{its context} determine the vector weights. BERT's setup and architecture have been shown to be general enough to be applicable to many natural language tasks. Importantly for Information Retrieval (IR), in contrast to prior deep learning solutions to IR problems which required significant tuning of neural net architectures and training regimes, "vanilla BERT" has been shown to outperform existing retrieval algorithms by a wide margin, including on tasks and corpora that have long resisted retrieval effectiveness gains over traditional IR baselines (such as Robust04). In this paper, we employ the recently proposed axiomatic dataset analysis technique -- that is, we create diagnostic datasets that each fulfil a retrieval heuristic (both term matching and semantic-based) -- to explore what BERT is able to learn. In contrast to our expectations, we find BERT, when applied to a recently released large-scale web corpus with ad-hoc topics, to \emph{not} adhere to any of the explored axioms. At the same time, BERT outperforms the traditional query likelihood retrieval model by 40\%. This means that the axiomatic approach to IR (and its extension of diagnostic datasets created for retrieval heuristics) may in its current form not be applicable to large-scale corpora. Additional -- different -- axioms are needed., Published at ECIR 2020