Many learning-to-rank (LtR) algorithms focus on query-independent model, in which query and document do not lie in the same feature space, and the rankers rely on the feature ensemble about query-document pair instead of the similarity between query instance and documents. However, existing algorithms do not consider local structures in query-document feature space, and are fragile to irrelevant noise features. In this paper, we propose a novel Riemannian metric learning algorithm to capture the local structures and develop a robust LtR algorithm. First, we design a concept called \textit{ideal candidate document} to introduce metric learning algorithm to query-independent model. Previous metric learning algorithms aiming to find an optimal metric space are only suitable for query-dependent model, in which query instance and documents belong to the same feature space and the similarity is directly computed from the metric space. Then we extend the new and extremely fast global Geometric Mean Metric Learning (GMML) algorithm to develop a localized GMML, namely L-GMML. Based on the combination of local learned metrics, we employ the popular Normalized Discounted Cumulative Gain~(NDCG) scorer and Weighted Approximate Rank Pairwise (WARP) loss to optimize the \textit{ideal candidate document} for each query candidate set. Finally, we can quickly evaluate all candidates via the similarity between the \textit{ideal candidate document} and other candidates. By leveraging the ability of metric learning algorithms to describe the complex structural information, our approach gives us a principled and efficient way to perform LtR tasks. The experiments on real-world datasets demonstrate that our proposed L-GMML algorithm outperforms the state-of-the-art metric learning to rank methods and the stylish query-independent LtR algorithms regarding accuracy and computational efficiency.
Detecting fine-grained differences in content conveyed in different languages matters for cross-lingual NLP and multilingual corpora analysis, but it is a challenging machine learning problem since annotation is expensive and hard to scale. This work improves the prediction and annotation of fine-grained semantic divergences. We introduce a training strategy for multilingual BERT models by learning to rank synthetic divergent examples of varying granularity. We evaluate our models on the Rationalized English-French Semantic Divergences, a new dataset released with this work, consisting of English-French sentence-pairs annotated with semantic divergence classes and token-level rationales. Learning to rank helps detect fine-grained sentence-level divergences more accurately than a strong sentence-level similarity model, while token-level predictions have the potential of further distinguishing between coarse and fine-grained divergences.
The task of session search focuses on using interaction data to improve relevance for the user's next query at the session level. In this paper, we formulate session search as a personalization task under the framework of learning to rank. Personalization approaches re-rank results to match a user model. Such user models are usually accumulated over time based on the user's browsing behaviour. We use a pre-computed and transparent set of user models based on concepts from the social science literature. Interaction data are used to map each session to these user models. Novel features are then estimated based on such models as well as sessions' interaction data. Extensive experiments on test collections from the TREC session track show statistically significant improvements over current session search algorithms.
Truth can vary over time. Fact-checking decisions on claim veracity should therefore take into account temporal information of both the claim and supporting or refuting evidence. In this work, we investigate the hypothesis that the timestamp of a Web page is crucial to how it should be ranked for a given claim. We delineate four temporal ranking methods that constrain evidence ranking differently and simulate hypothesis-specific evidence rankings given the evidence timestamps as gold standard. Evidence ranking in three fact-checking models is ultimately optimized using a learning-to-rank loss function. Our study reveals that time-aware evidence ranking not only surpasses relevance assumptions based purely on semantic similarity or position in a search results list, but also improves veracity predictions of time-sensitive claims in particular.
Learning the optimal ordering of content is an important challenge in website design. The learning to rank (LTR) framework models this problem as a sequential problem of selecting lists of content and observing where users decide to click. Most previous work on LTR assumes that the user considers each item in the list in isolation, and makes binary choices to click or not on each. We introduce a multinomial logit (MNL) choice model to the LTR framework, which captures the behaviour of users who consider the ordered list of items as a whole and make a single choice among all the items and a no-click option. Under the MNL model, the user favours items which are either inherently more attractive, or placed in a preferable position within the list. We propose upper confidence bound (UCB) algorithms to minimise regret in two settings - where the position dependent parameters are known, and unknown. We present theoretical analysis leading to an $\Omega(\sqrt{JT})$ lower bound for the problem, an $\tilde{O}(\sqrt{JT})$ upper bound on regret of the UCB algorithm in the known-parameter setting, and an $\tilde{O}(K^2\sqrt{JT})$ upper bound on regret, the first, in the more challenging unknown-position-parameter setting. Our analyses are based on tight new concentration results for Geometric random variables, and novel functional inequalities for maximum likelihood estimators computed on discrete data.
Neural ranking models for information retrieval (IR) use shallow or deep neural networks to rank search results in response to a query. Traditional learning to rank models employ machine learning techniques over hand-crafted IR features. By contrast, neural models learn representations of language from raw text that can bridge the gap between query and document vocabulary. Unlike classical IR models, these new machine learning based approaches are data-hungry, requiring large scale training data before they can be deployed. This tutorial introduces basic concepts and intuitions behind neural IR models, and places them in the context of traditional retrieval models. We begin by introducing fundamental concepts of IR and different neural and non-neural approaches to learning vector representations of text. We then review shallow neural IR methods that employ pre-trained neural term embeddings without learning the IR task end-to-end. We introduce deep neural networks next, discussing popular deep architectures. Finally, we review the current DNN models for information retrieval. We conclude with a discussion on potential future directions for neural IR.
Learning effective feature crosses is the key behind building recommender systems. However, the sparse and large feature space requires exhaustive search to identify effective crosses. Deep & Cross Network (DCN) was proposed to automatically and efficiently learn bounded-degree predictive feature interactions. Unfortunately, in models that serve web-scale traffic with billions of training examples, DCN showed limited expressiveness in its cross network at learning more predictive feature interactions. Despite significant research progress made, many deep learning models in production still rely on traditional feed-forward neural networks to learn feature crosses inefficiently. In light of the pros/cons of DCN and existing feature interaction learning approaches, we propose an improved framework DCN-V2 to make DCN more practical in large-scale industrial settings. In a comprehensive experimental study with extensive hyper-parameter search and model tuning, we observed that DCN-V2 approaches outperform all the state-of-the-art algorithms on popular benchmark datasets. The improved DCN-V2 is more expressive yet remains cost efficient at feature interaction learning, especially when coupled with a mixture of low-rank architecture. DCN-V2 is simple, can be easily adopted as building blocks, and has delivered significant offline accuracy and online business metrics gains across many web-scale learning to rank systems at Google.
Learning-to-rank has been intensively studied and has shown significantly increasing values in a wide range of domains. The performance of learning-to-rank methods is commonly evaluated using rank-sensitive metrics, such as average precision (AP) and normalized discounted cumulative gain (nDCG). Unfortunately, how to effectively optimize rank-sensitive objectives is far from being resolved, which has been an open problem since the dawn of learning-to-rank over a decade ago. In this paper, we introduce a simple yet effective framework for directly optimizing information retrieval (IR) metrics. Specifically, we propose a novel twin-sigmoid function for deriving the exact rank positions of documents during the optimization process instead of using approximated rank positions or relying on the traditional sorting algorithms. Thanks to this, the rank positions are differentiable, enabling us to reformulate the widely used IR metrics as differentiable ones and directly optimize them based on neural networks. Furthermore, by carrying out an in-depth analysis of the gradients, we pinpoint the potential limitations inherent with direct optimization of IR metrics based on the vanilla sigmoid. To break the limitations, we propose different strategies by explicitly modifying the gradient computation. To validate the effectiveness of the proposed framework for direct optimization of IR metrics, we conduct a series of experiments on the widely used benchmark collection MSLRWEB30K.
Deep neural networks has become the first choice for researchers working on algorithmic aspects of learning-to-rank. Unfortunately, it is not trivial to find the optimal setting of hyper-parameters that achieves the best ranking performance. As a result, it becomes more and more difficult to develop a new model and conduct a fair comparison with prior methods, especially for newcomers. In this work, we propose PT-Ranking, an open-source project based on PyTorch for developing and evaluating learning-to-rank methods using deep neural networks as the basis to construct a scoring function. On one hand, PT-Ranking includes many representative learning-to-rank methods. Besides the traditional optimization framework via empirical risk minimization, adversarial optimization framework is also integrated. Furthermore, PT-Ranking's modular design provides a set of building blocks that users can leverage to develop new ranking models. On the other hand, PT-Ranking supports to compare different learning-to-rank methods based on the widely used datasets (e.g., MSLR-WEB30K, Yahoo!LETOR and Istella LETOR) in terms of different metrics, such as precision, MAP, nDCG, nERR. By randomly masking the ground-truth labels with a specified ratio, PT-Ranking allows to examine to what extent the ratio of unlabelled query-document pairs affects the performance of different learning-to-rank methods. We further conducted a series of demo experiments to clearly show the effect of different factors on neural learning-to-rank methods, such as the activation function, the number of layers and the optimization strategy.
Online food ordering marketplaces are multi-stakeholder systems where recommendations impact the experience and growth of each participant in the system. A recommender system in this setting has to encapsulate the objectives and constraints of different stakeholders in order to find utility of an item for recommendation. Constrained-optimization based approaches to this problem typically involve complex formulations and have high computational complexity in production settings involving millions of entities. Simplifications and relaxation techniques (for example, scalarization) help but introduce sub-optimality and can be time-consuming due to the amount of tuning needed. In this paper, we introduce a method involving multi-goal sampling followed by ranking for user-relevance (Sample-Rank), to nudge recommendations towards multi-objective (MO) goals of the marketplace. The proposed method's novelty is that it reduces the MO recommendation problem to sampling from a desired multi-goal distribution then using it to build a production-friendly learning-to-rank (LTR) model. In offline experiments we show that we are able to bias recommendations towards MO criteria with acceptable trade-offs in metrics like AUC and NDCG. We also show results from a large-scale online A/B experiment where this approach gave a statistically significant lift of 2.64% in average revenue per order (RPO) (objective #1) with no drop in conversion rate (CR) (objective #2) while holding the average last-mile traversed flat (objective #3), vs. the baseline ranking method. This method also significantly reduces time to model development and deployment in MO settings and allows for trivial extensions to more objectives and other types of LTR models.
Besides position bias, which has been well-studied, trust bias is another type of bias prevalent in user interactions with rankings: users are more likely to click incorrectly w.r.t. their preferences on highly ranked items because they trust the ranking system. While previous work has observed this behavior in users, we prove that existing Counterfactual Learning to Rank (CLTR) methods do not remove this bias, including methods specifically designed to mitigate this type of bias. Moreover, we prove that Inverse Propensity Scoring (IPS) is principally unable to correct for trust bias under non-trivial circumstances. Our main contribution is a new estimator based on affine corrections: it both reweights clicks and penalizes items displayed on ranks with high trust bias. Our estimator is the first estimator that is proven to remove the effect of both trust bias and position bias. Furthermore, we show that our estimator is a generalization of the existing CLTR framework: if no trust bias is present, it reduces to the original IPS estimator. Our semi-synthetic experiments indicate that by removing the effect of trust bias in addition to position bias, CLTR can approximate the optimal ranking system even closer than previously possible.
Online Learning to Rank (OLTR) methods optimize rankers based on user interactions. State-of-the-art OLTR methods are built specifically for linear models. Their approaches do not extend well to non-linear models such as neural networks. We introduce an entirely novel approach to OLTR that constructs a weighted differentiable pairwise loss after each interaction: Pairwise Differentiable Gradient Descent (PDGD). PDGD breaks away from the traditional approach that relies on interleaving or multileaving and extensive sampling of models to estimate gradients. Instead, its gradient is based on inferring preferences between document pairs from user clicks and can optimize any differentiable model. We prove that the gradient of PDGD is unbiased w.r.t. user document pair preferences. Our experiments on the largest publicly available Learning to Rank (LTR) datasets show considerable and significant improvements under all levels of interaction noise. PDGD outperforms existing OLTR methods both in terms of learning speed as well as final convergence. Furthermore, unlike previous OLTR methods, PDGD also allows for non-linear models to be optimized effectively. Our results show that using a neural network leads to even better performance at convergence than a linear model. In summary, PDGD is an efficient and unbiased OLTR approach that provides a better user experience than previously possible.
Learning to rank has been intensively studied and widely applied in information retrieval. Typically, a global ranking function is learned from a set of labeled data, which can achieve good performance on average but may be suboptimal for individual queries by ignoring the fact that relevant documents for different queries may have different distributions in the feature space. Inspired by the idea of pseudo relevance feedback where top ranked documents, which we refer as the \textit{local ranking context}, can provide important information about the query's characteristics, we propose to use the inherent feature distributions of the top results to learn a Deep Listwise Context Model that helps us fine tune the initial ranked list. Specifically, we employ a recurrent neural network to sequentially encode the top results using their feature vectors, learn a local context model and use it to re-rank the top results. There are three merits with our model: (1) Our model can capture the local ranking context based on the complex interactions between top results using a deep neural network; (2) Our model can be built upon existing learning-to-rank methods by directly using their extracted feature vectors; (3) Our model is trained with an attention-based loss function, which is more effective and efficient than many existing listwise methods. Experimental results show that the proposed model can significantly improve the state-of-the-art learning to rank methods on benchmark retrieval corpora.
Leveraging biased click data for optimizing learning to rank systems has been a popular approach in information retrieval. Because click data is often noisy and biased, a variety of methods have been proposed to construct unbiased learning to rank (ULTR) algorithms for the learning of unbiased ranking models. Among them, automatic unbiased learning to rank (AutoULTR) algorithms that jointly learn user bias models (i.e., propensity models) with unbiased rankers have received a lot of attention due to their superior performance and low deployment cost in practice. Despite their differences in theories and algorithm design, existing studies on ULTR usually use uni-variate ranking functions to score each document or result independently. On the other hand, recent advances in context-aware learning-to-rank models have shown that multivariate scoring functions, which read multiple documents together and predict their ranking scores jointly, are more powerful than uni-variate ranking functions in ranking tasks with human-annotated relevance labels. Whether such superior performance would hold in ULTR with noisy data, however, is mostly unknown. In this paper, we investigate existing multivariate scoring functions and AutoULTR algorithms in theory and prove that permutation invariance is a crucial factor that determines whether a context-aware learning-to-rank model could be applied to existing AutoULTR framework. Our experiments with synthetic clicks on two large-scale benchmark datasets show that AutoULTR models with permutation-invariant multivariate scoring functions significantly outperform those with uni-variate scoring functions and permutation-variant multivariate scoring functions.
In this paper, we quest the capability of transferring the quality of natural scene images to the images that are not acquired by optical cameras (e.g., screen content images, SCIs), rooted in the widely accepted view that the human visual system has adapted and evolved through the perception of natural environment. Here, we develop the first unsupervised domain adaptation based no reference quality assessment method for SCIs, leveraging rich subjective ratings of the natural images (NIs). In general, it is a non-trivial task to directly transfer the quality prediction model from NIs to a new type of content (i.e., SCIs) that holds dramatically different statistical characteristics. Inspired by the transferability of pair-wise relationship, the proposed quality measure operates based on the philosophy of improving the transferability and discriminability simultaneously. In particular, we introduce three types of losses which complementarily and explicitly regularize the feature space of ranking in a progressive manner. Regarding feature discriminatory capability enhancement, we propose a center based loss to rectify the classifier and improve its prediction capability not only for source domain (NI) but also the target domain (SCI). For feature discrepancy minimization, the maximum mean discrepancy (MMD) is imposed on the extracted ranking features of NIs and SCIs. Furthermore, to further enhance the feature diversity, we introduce the correlation penalization between different feature dimensions, leading to the features with lower rank and higher diversity. Experiments show that our method can achieve higher performance on different source-target settings based on a light-weight convolution neural network. The proposed method also sheds light on learning quality assessment measures for unseen application-specific content without the cumbersome and costing subjective evaluations.
The animated Graphical Interchange Format (GIF) images have been widely used on social media as an intuitive way of expression emotion. Given their expressiveness, GIFs offer a more nuanced and precise way to convey emotions. In this paper, we present our solution for the EmotionGIF 2020 challenge, the shared task of SocialNLP 2020. To recommend GIF categories for unlabeled tweets, we regarded this problem as a kind of matching tasks and proposed a learning to rank framework based on Bidirectional Encoder Representations from Transformer (BERT) and LightGBM. Our team won the 4th place with a Mean Average Precision @ 6 (MAP@6) score of 0.5394 on the round 1 leaderboard.
Recommendation services are extensively adopted in several user-centered applications as a tool to alleviate the information overload problem and help users in orienteering in a vast space of possible choices. In such scenarios, data ownership is a crucial concern since users may not be willing to share their sensitive preferences (e.g., visited locations) with a central server. Unfortunately, data harvesting and collection is at the basis of modern, state-of-the-art approaches to recommendation. To address this issue, we present FPL, an architecture in which users collaborate in training a central factorization model while controlling the amount of sensitive data leaving their devices. The proposed approach implements pair-wise learning-to-rank optimization by following the Federated Learning principles, originally conceived to mitigate the privacy risks of traditional machine learning. The public implementation is available at https://split.to/sisinflab-fpl.
Active learning emerged as an alternative to alleviate the effort to label huge amount of data for data hungry applications (such as image/video indexing and retrieval, autonomous driving, etc.). The goal of active learning is to automatically select a number of unlabeled samples for annotation (according to a budget), based on an acquisition function, which indicates how valuable a sample is for training the model. The learning loss method is a task-agnostic approach which attaches a module to learn to predict the target loss of unlabeled data, and select data with the highest loss for labeling. In this work, we follow this strategy but we define the acquisition function as a learning to rank problem and rethink the structure of the loss prediction module, using a simple but effective listwise approach. Experimental results on four datasets demonstrate that our method outperforms recent state-of-the-art active learning approaches for both image classification and regression tasks.
Axis-aligned decision forests have long been the leading class of machine learning algorithms for modeling tabular data. In many applications of machine learning such as learning-to-rank, decision forests deliver remarkable performance. They also possess other coveted characteristics such as interpretability. Despite their widespread use and rich history, decision forests to date fail to consume raw structured data such as text, or learn effective representations for them, a factor behind the success of deep neural networks in recent years. While there exist methods that construct smoothed decision forests to achieve representation learning, the resulting models are decision forests in name only: They are no longer axis-aligned, use stochastic decisions, or are not interpretable. Furthermore, none of the existing methods are appropriate for problems that require a Transfer Learning treatment. In this work, we present a novel but intuitive proposal to achieve representation learning for decision forests without imposing new restrictions or necessitating structural changes. Our model is simply a decision forest, possibly trained using any forest learning algorithm, atop a deep neural network. By approximating the gradients of the decision forest through input perturbation, a purely analytical procedure, the decision forest directs the neural network to learn or fine-tune representations. Our framework has the advantage that it is applicable to any arbitrary decision forest and that it allows the use of arbitrary deep neural networks for representation learning. We demonstrate the feasibility and effectiveness of our proposal through experiments on synthetic and benchmark classification datasets.
Background:Cognitive assessments represent the most common clinical routine for the diagnosis of Alzheimer's Disease (AD). Given a large number of cognitive assessment tools and time-limited office visits, it is important to determine a proper set of cognitive tests for different subjects. Most current studies create guidelines of cognitive test selection for a targeted population, but they are not customized for each individual subject. In this manuscript, we develop a machine learning paradigm enabling personalized cognitive assessments prioritization. Method: We adapt a newly developed learning-to-rank approach PLTR to implement our paradigm. This method learns the latent scoring function that pushes the most effective cognitive assessments onto the top of the prioritization list. We also extend PLTR to better separate the most effective cognitive assessments and the less effective ones. Results: Our empirical study on the ADNI data shows that the proposed paradigm outperforms the state-of-the-art baselines on identifying and prioritizing individual-specific cognitive biomarkers. We conduct experiments in cross validation and level-out validation settings. In the two settings, our paradigm significantly outperforms the best baselines with improvement as much as 22.1% and 19.7%, respectively, on prioritizing cognitive features. Conclusions: The proposed paradigm achieves superior performance on prioritizing cognitive biomarkers. The cognitive biomarkers prioritized on top have great potentials to facilitate personalized diagnosis, disease subtyping, and ultimately precision medicine in AD.