arXiv subjects

Maarten de Rijke

Publications and source records attributed to Maarten de Rijke.

At least 253 records · Page 14Linked to original sources

arXiv2019

Thinking Globally, Acting Locally: Distantly Supervised Global-to-Local Knowledge Selection for Background Based Conversation

Background Based Conversations (BBCs) have been introduced to help conversational systems avoid generating overly generic responses. In a BBC, the conversation is grounded in a knowledge source. A key challenge in BBCs is Knowledge Selection (KS): given a conversational context, try to find the appropriate background knowledge (a text fragment containing related facts or comments, etc.) based on which to generate the next response. Previous work addresses KS by employing attention and/or pointer mechanisms. These mechanisms use a local perspective, i.e., they select a token at a time based solely on the current decoding state. We argue for the adoption of a global perspective, i.e., pre-selecting some text fragments from the background knowledge that could help determine the topic of the next response. We enhance KS in BBCs by introducing a Global-to-Local Knowledge Selection (GLKS) mechanism. Given a conversational context and background knowledge, we first learn a topic transition vector to encode the most likely text fragments to be used in the next response, which is then used to guide the local KS at each decoding timestamp. In order to effectively learn the topic transition vector, we propose a distantly supervised learning schema. Experimental results show that the GLKS model significantly outperforms state-of-the-art methods in terms of both automatic and human evaluation. More importantly, GLKS achieves this without requiring any extra annotations, which demonstrates its high degree of scalability.

cs.CL↗

arXiv2019

Understanding Multi-Head Attention in Abstractive Summarization

Attention mechanisms in deep learning architectures have often been used as a means of transparency and, as such, to shed light on the inner workings of the architectures. Recently, there has been a growing interest in whether or not this assumption is correct. In this paper we investigate the interpretability of multi-head attention in abstractive summarization, a sequence-to-sequence task for which attention does not have an intuitive alignment role, such as in machine translation. We first introduce three metrics to gain insight in the focus of attention heads and observe that these heads specialize towards relative positions, specific part-of-speech tags, and named entities. However, we also find that ablating and pruning these heads does not lead to a significant drop in performance, indicating redundancy. By replacing the softmax activation functions with sparsemax activation functions, we find that attention heads behave seemingly more transparent: we can ablate fewer heads and heads score higher on our interpretability metrics. However, if we apply pruning to the sparsemax model we find that we can prune even more heads, raising the question whether enforced sparsity actually improves transparency. Finally, we find that relative positions heads seem integral to summarization performance and persistently remain after pruning.

cs.CL↗

arXiv2019

Pre-train, Interact, Fine-tune: A Novel Interaction Representation for Text Classification

Text representation can aid machines in understanding text. Previous work on text representation often focuses on the so-called forward implication, i.e., preceding words are taken as the context of later words for creating representations, thus ignoring the fact that the semantics of a text segment is a product of the mutual implication of words in the text: later words contribute to the meaning of preceding words. We introduce the concept of interaction and propose a two-perspective interaction representation, that encapsulates a local and a global interaction representation. Here, a local interaction representation is one that interacts among words with parent-children relationships on the syntactic trees and a global interaction interpretation is one that interacts among all the words in a sentence. We combine the two interaction representations to develop a Hybrid Interaction Representation (HIR). Inspired by existing feature-based and fine-tuning-based pretrain-finetuning approaches to language models, we integrate the advantages of feature-based and fine-tuning-based methods to propose the Pre-train, Interact, Fine-tune (PIF) architecture. We evaluate our proposed models on five widely-used datasets for text classification tasks. Our ensemble method, outperforms state-of-the-art baselines with improvements ranging from 2.03% to 3.15% in terms of error rate. In addition, we find that, the improvements of PIF against most state-of-the-art methods is not affected by increasing of the length of the text.

cs.IR↗

arXiv2019

SEntNet: Source-aware Recurrent Entity Network for Dialogue Response Selection

Dialogue response selection is an important part of Task-oriented Dialogue Systems (TDSs); it aims to predict an appropriate response given a dialogue context. Obtaining key information from a complex, long dialogue context is challenging, especially when different sources of information are available, e.g., the user's utterances, the system's responses, and results retrieved from a knowledge base (KB). Previous work ignores the type of information source and merges sources for response selection. However, accounting for the source type may lead to remarkable differences in the quality of response selection. We propose the Source-aware Recurrent Entity Network (SEntNet), which is aware of different information sources for the response selection process. SEntNet achieves this by employing source-specific memories to exploit differences in the usage of words and syntactic structure from different information sources (user, system, and KB). Experimental results show that SEntNet obtains 91.0% accuracy on the Dialog bAbI dataset, outperforming prior work by 4.7%. On the DSTC2 dataset, SEntNet obtains an accuracy of 41.2%, beating source unaware recurrent entity networks by 2.4%.

cs.IR↗

arXiv2019

Improving End-to-End Sequential Recommendations with Intent-aware Diversification

Sequential Recommendation (SRs) that capture users' dynamic intents by modeling user sequential behaviors can recommend closely accurate products to users. Previous work on SRs is mostly focused on optimizing the recommendation accuracy, often ignoring the recommendation diversity, even though it is an important criterion for evaluating the recommendation performance. Most existing methods for improving the diversity of recommendations are not ideally applicable for SRs because they assume that user intents are static and rely on post-processing the list of recommendations to promote diversity. We consider both recommendation accuracy and diversity for SRs by proposing an end-to-end neural model, called Intent-aware Diversified Sequential Recommendation (IDSR). Specifically, we introduce an Implicit Intent Mining module (IIM) into SRs to capture different user intents reflected in user behavior sequences. Then, we design an Intent-aware Diversity Promoting (IDP) loss to supervise the learning of the IIM module and force the model to take recommendation diversity into consideration during training. Extensive experiments on two benchmark datasets show that IDSR significantly outperforms state-of-the-art methods in terms of recommendation diversity while yielding comparable or superior recommendation accuracy.

cs.IR↗

arXiv2019

Improving Outfit Recommendation with Co-supervision of Fashion Generation

The task of fashion recommendation includes two main challenges: visual understanding and visual matching. Visual understanding aims to extract effective visual features. Visual matching aims to model a human notion of compatibility to compute a match between fashion items. Most previous studies rely on recommendation loss alone to guide visual understanding and matching. Although the features captured by these methods describe basic characteristics (e.g., color, texture, shape) of the input items, they are not directly related to the visual signals of the output items (to be recommended). This is problematic because the aesthetic characteristics (e.g., style, design), based on which we can directly infer the output items, are lacking. Features are learned under the recommendation loss alone, where the supervision signal is simply whether the given two items are matched or not. To address this problem, we propose a neural co-supervision learning framework, called the FAshion Recommendation Machine (FARM). FARM improves visual understanding by incorporating the supervision of generation loss, which we hypothesize to be able to better encode aesthetic information. FARM enhances visual matching by introducing a novel layer-to-layer matching mechanism to fuse aesthetic information more effectively, and meanwhile avoiding paying too much attention to the generation quality and ignoring the recommendation performance. Extensive experiments on two publicly available datasets show that FARM outperforms state-of-the-art models on outfit recommendation, in terms of AUC and MRR. Detailed analyses of generated and recommended items demonstrate that FARM can encode better features and generate high quality images as references to improve recommendation performance.

cs.IR↗

arXiv2019

Message Passing for Complex Question Answering over Knowledge Graphs

Question answering over knowledge graphs (KGQA) has evolved from simple single-fact questions to complex questions that require graph traversal and aggregation. We propose a novel approach for complex KGQA that uses unsupervised message passing, which propagates confidence scores obtained by parsing an input question and matching terms in the knowledge graph to a set of possible answers. First, we identify entity, relationship, and class names mentioned in a natural language question, and map these to their counterparts in the graph. Then, the confidence scores of these mappings propagate through the graph structure to locate the answer entities. Finally, these are aggregated depending on the identified question type. This approach can be efficiently implemented as a series of sparse matrix multiplications mimicking joins over small local subgraphs. Our evaluation results show that the proposed approach outperforms the state-of-the-art on the LC-QuAD benchmark. Moreover, we show that the performance of the approach depends only on the quality of the question interpretation results, i.e., given a correct relevance score distribution, our approach always produces a correct answer ranking. Our error analysis reveals correct answers missing from the benchmark dataset and inconsistencies in the DBpedia knowledge graph. Finally, we provide a comprehensive evaluation of the proposed approach accompanied with an ablation study and an error analysis, which showcase the pitfalls for each of the question answering components in more detail.

cs.CL↗

arXiv2019

Unbiased Learning to Rank: Counterfactual and Online Approaches

This tutorial covers and contrasts the two main methodologies in unbiased Learning to Rank (LTR): Counterfactual LTR and Online LTR. There has long been an interest in LTR from user interactions, however, this form of implicit feedback is very biased. In recent years, unbiased LTR methods have been introduced to remove the effect of different types of bias caused by user-behavior in search. For instance, a well addressed type of bias is position bias: the rank at which a document is displayed heavily affects the interactions it receives. Counterfactual LTR methods deal with such types of bias by learning from historical interactions while correcting for the effect of the explicitly modelled biases. Online LTR does not use an explicit user model, in contrast, it learns through an interactive process where randomized results are displayed to the user. Through randomization the effect of different types of bias can be removed from the learning process. Though both methodologies lead to unbiased LTR, their approaches differ considerably, furthermore, so do their theoretical guarantees, empirical results, effects on the user experience during learning, and applicability. Consequently, for practitioners the choice between the two is very substantial. By providing an overview of both approaches and contrasting them, we aim to provide an essential guide to unbiased LTR so as to aid in understanding and choosing between methodologies.

cs.IR↗

arXiv2019

To Model or to Intervene: A Comparison of Counterfactual and Online Learning to Rank from User Interactions

Learning to Rank (LTR) from user interactions is challenging as user feedback often contains high levels of bias and noise. At the moment, two methodologies for dealing with bias prevail in the field of LTR: counterfactual methods that learn from historical data and model user behavior to deal with biases; and online methods that perform interventions to deal with bias but use no explicit user models. For practitioners the decision between either methodology is very important because of its direct impact on end users. Nevertheless, there has never been a direct comparison between these two approaches to unbiased LTR. In this study we provide the first benchmarking of both counterfactual and online LTR methods under different experimental conditions. Our results show that the choice between the methodologies is consequential and depends on the presence of selection bias, and the degree of position bias and interaction noise. In settings with little bias or noise counterfactual methods can obtain the highest ranking performance; however, in other circumstances their optimization can be detrimental to the user experience. Conversely, online methods are very robust to bias and noise but require control over the displayed rankings. Our findings confirm and contradict existing expectations on the impact of model-based and intervention-based methods in LTR, and allow practitioners to make an informed decision between the two methodologies.

cs.IR↗

arXiv2019

Proceedings of FACTS-IR 2019

The proceedings list for the program of FACTS-IR 2019, the Workshop on Fairness, Accountability, Confidentiality, Transparency, and Safety in Information Retrieval held at SIGIR 2019.

cs.IR↗

arXiv2019

Joint Neural Collaborative Filtering for Recommender Systems

We propose a J-NCF method for recommender systems. The J-NCF model applies a joint neural network that couples deep feature learning and deep interaction modeling with a rating matrix. Deep feature learning extracts feature representations of users and items with a deep learning architecture based on a user-item rating matrix. Deep interaction modeling captures non-linear user-item interactions with a deep neural network using the feature representations generated by the deep feature learning process as input. J-NCF enables the deep feature learning and deep interaction modeling processes to optimize each other through joint training, which leads to improved recommendation performance. In addition, we design a new loss function for optimization, which takes both implicit and explicit feedback, point-wise and pair-wise loss into account. Experiments on several real-word datasets show significant improvements of J-NCF over state-of-the-art methods, with improvements of up to 8.24% on the MovieLens 100K dataset, 10.81% on the MovieLens 1M dataset, and 10.21% on the Amazon Movies dataset in terms of HR@10. NDCG@10 improvements are 12.42%, 14.24% and 15.06%, respectively. We also conduct experiments to evaluate the scalability and sensitivity of J-NCF. Our experiments show that the J-NCF model has a competitive recommendation performance with inactive users and different degrees of data sparsity when compared to state-of-the-art baselines.

cs.IR↗

arXiv2019

A Modular Task-oriented Dialogue System Using a Neural Mixture-of-Experts

End-to-end Task-oriented Dialogue Systems (TDSs) have attracted a lot of attention for their superiority (e.g., in terms of global optimization) over pipeline modularized TDSs. Previous studies on end-to-end TDSs use a single-module model to generate responses for complex dialogue contexts. However, no model consistently outperforms the others in all cases. We propose a neural Modular Task-oriented Dialogue System(MTDS) framework, in which a few expert bots are combined to generate the response for a given dialogue context. MTDS consists of a chair bot and several expert bots. Each expert bot is specialized for a particular situation, e.g., one domain, one type of action of a system, etc. The chair bot coordinates multiple expert bots and adaptively selects an expert bot to generate the appropriate response. We further propose a Token-level Mixture-of-Expert (TokenMoE) model to implement MTDS, where the expert bots predict multiple tokens at each timestamp and the chair bot determines the final generated token by fully taking into consideration the outputs of all expert bots. Both the chair bot and the expert bots are jointly trained in an end-to-end fashion. To verify the effectiveness of TokenMoE, we carry out extensive experiments on a benchmark dataset. Compared with the baseline using a single-module model, our TokenMoE improves the performance by 8.1% of inform rate and 0.8% of success rate.

cs.CL↗

arXiv2019

Do Transformer Attention Heads Provide Transparency in Abstractive Summarization?

Learning algorithms become more powerful, often at the cost of increased complexity. In response, the demand for algorithms to be transparent is growing. In NLP tasks, attention distributions learned by attention-based deep learning models are used to gain insights in the models' behavior. To which extent is this perspective valid for all NLP tasks? We investigate whether distributions calculated by different attention heads in a transformer architecture can be used to improve transparency in the task of abstractive summarization. To this end, we present both a qualitative and quantitative analysis to investigate the behavior of the attention heads. We show that some attention heads indeed specialize towards syntactically and semantically distinct input. We propose an approach to evaluate to which extent the Transformer model relies on specifically learned attention distributions. We also discuss what this implies for using attention distributions as a means of transparency.

cs.CL↗

arXiv2019

Explaining Predictions from Tree-based Boosting Ensembles

Understanding how "black-box" models arrive at their predictions has sparked significant interest from both within and outside the AI community. Our work focuses on doing this by generating local explanations about individual predictions for tree-based ensembles, specifically Gradient Boosting Decision Trees (GBDTs). Given a correctly predicted instance in the training set, we wish to generate a counterfactual explanation for this instance, that is, the minimal perturbation of this instance such that the prediction flips to the opposite class. Most existing methods for counterfactual explanations are (1) model-agnostic, so they do not take into account the structure of the original model, and/or (2) involve building a surrogate model on top of the original model, which is not guaranteed to represent the original model accurately. There exists a method specifically for random forests; we wish to extend this method for GBDTs. This involves accounting for (1) the sequential dependency between trees and (2) training on the negative gradients instead of the original labels.

cs.LG↗

arXiv2019

BubbleRank: Safe Online Learning to Re-Rank via Implicit Click Feedback

In this paper, we study the problem of safe online learning to re-rank, where user feedback is used to improve the quality of displayed lists. Learning to rank has traditionally been studied in two settings. In the offline setting, rankers are typically learned from relevance labels created by judges. This approach has generally become standard in industrial applications of ranking, such as search. However, this approach lacks exploration and thus is limited by the information content of the offline training data. In the online setting, an algorithm can experiment with lists and learn from feedback on them in a sequential fashion. Bandit algorithms are well-suited for this setting but they tend to learn user preferences from scratch, which results in a high initial cost of exploration. This poses an additional challenge of safe exploration in ranked lists. We propose BubbleRank, a bandit algorithm for safe re-ranking that combines the strengths of both the offline and online settings. The algorithm starts with an initial base list and improves it online by gradually exchanging higher-ranked less attractive items for lower-ranked more attractive items. We prove an upper bound on the n-step regret of BubbleRank that degrades gracefully with the quality of the initial base list. Our theoretical findings are supported by extensive experiments on a large-scale real-world click dataset.

cs.LG↗

arXiv2019

Improving Background Based Conversation with Context-aware Knowledge Pre-selection

Background Based Conversations (BBCs) have been developed to make dialogue systems generate more informative and natural responses by leveraging background knowledge. Existing methods for BBCs can be grouped into two categories: extraction-based methods and generation-based methods. The former extract spans frombackground material as responses that are not necessarily natural. The latter generate responses thatare natural but not necessarily effective in leveraging background knowledge. In this paper, we focus on generation-based methods and propose a model, namely Context-aware Knowledge Pre-selection (CaKe), which introduces a pre-selection process that uses dynamic bi-directional attention to improve knowledge selection by using the utterance history context as prior information to select the most relevant background material. Experimental results show that our model is superior to current state-of-the-art baselines, indicating that it benefits from the pre-selection process, thus improving in-formativeness and fluency.

cs.CL↗

arXiv2019

ViTOR: Learning to Rank Webpages Based on Visual Features

The visual appearance of a webpage carries valuable information about its quality and can be used to improve the performance of learning to rank (LTR). We introduce the Visual learning TO Rank (ViTOR) model that integrates state-of-the-art visual features extraction methods by (i) transfer learning from a pre-trained image classification model, and (ii) synthetic saliency heat maps generated from webpage snapshots. Since there is currently no public dataset for the task of LTR with visual features, we also introduce and release the ViTOR dataset, containing visually rich and diverse webpages. The ViTOR dataset consists of visual snapshots, non-visual features and relevance judgments for ClueWeb12 webpages and TREC Web Track queries. We experiment with the proposed ViTOR model on the ViTOR dataset and show that it significantly improves the performance of LTR with visual features

cs.IR↗

arXiv2019

Explainable Outfit Recommendation with Joint Outfit Matching and Comment Generation

Most previous work on outfit recommendation focuses on designing visual features to enhance recommendations. Existing work neglects user comments of fashion items, which have been proved to be effective in generating explanations along with better recommendation results. We propose a novel neural network framework, neural outfit recommendation (NOR), that simultaneously provides outfit recommendations and generates abstractive comments. NOR consists of two parts: outfit matching and comment generation. For outfit matching, we propose a convolutional neural network with a mutual attention mechanism to extract visual features. The visual features are then decoded into a rating score for the matching prediction. For abstractive comment generation, we propose a gated recurrent neural network with a cross-modality attention mechanism to transform visual features into a concise sentence. The two parts are jointly trained based on a multi-task learning framework in an end-to-end back-propagation paradigm. Extensive experiments conducted on an existing dataset and a collected real-world dataset show NOR achieves significant improvements over state-of-the-art baselines for outfit recommendation. Meanwhile, our generated comments achieve impressive ROUGE and BLEU scores in comparison to human-written comments. The generated comments can be regarded as explanations for the recommendation results. We release the dataset and code to facilitate future research.

cs.IR↗