Neural generation methods for task-oriented dialogue typically generate from a meaning representation that is populated using a database of domain information, such as a table of data describing a restaurant. While earlier work focused solely on the semantic fidelity of outputs, recent work has started to explore methods for controlling the style of the generated text while simultaneously achieving semantic accuracy. Here we experiment with two stylistic benchmark tasks, generating language that exhibits variation in personality, and generating discourse contrast. We report a huge performance improvement in both stylistic control and semantic accuracy over the state of the art on both of these benchmarks. We test several different models and show that putting stylistic conditioning in the decoder and eliminating the semantic re-ranker used in earlier models results in more than 15 points higher BLEU for Personality, with a reduction of semantic error to near zero. We also report an improvement from .75 to .81 in controlling contrast and a reduction in semantic error from 16% to 2%.

Previous work on visual storytelling mainly focused on exploring image sequence as evidence for storytelling and neglected textual evidence for guiding story generation. Motivated by human storytelling process which recalls stories for familiar images, we exploit textual evidence from similar images to help generate coherent and meaningful stories. To pick the images which may provide textual experience, we propose a two-step ranking method based on image object recognition techniques. To utilize textual information, we design an extended Seq2Seq model with two-channel encoder and attention. Experiments on the VIST dataset show that our method outperforms state-of-the-art baseline models without heavy engineering.

The move from pipeline Natural Language Generation (NLG) approaches to neural end-to-end approaches led to a loss of control in sentence planning operations owing to the conflation of intermediary micro-planning stages into a single model. Such control is highly necessary when the text should be tailored to respect some constraints such as which entity to be mentioned first, the entity position, the complexity of sentences, etc. In this paper, we introduce fine-grained control of sentence planning in neural data-to-text generation models at two levels - realization of input entities in desired sentences and realization of the input entities in the desired position among individual sentences. We show that by augmenting the input with explicit position identifiers, the neural model can achieve a great control over the output structure while keeping the naturalness of the generated text intact. Since sentence level metrics are not entirely suitable to evaluate this task, we used a metric specific to our task that accounts for the model’s ability to achieve control. The results demonstrate that the position identifiers do constraint the neural model to respect the intended output structure which can be useful in a variety of domains that require the generated text to be in a certain structure.

Due to the absence of labeled data, discourse parsing still remains challenging in some languages. In this paper, we present a simple and efficient method to conduct zero-shot Chinese text-level dependency parsing by leveraging English discourse labeled data and parsing techniques. We first construct the Chinese-English mapping from the level of sentence and elementary discourse unit (EDU), and then exploit the parsing results of the corresponding English translations to obtain the discourse trees for the Chinese text. This method can automatically conduct Chinese discourse parsing, with no need of a large scale of Chinese labeled data.

When generating technical instructions, it is often necessary to describe an object that does not exist yet. For example, an NLG system which explains how to build a house needs to generate sentences like “build *a wall of height five to your left*” and “now build *a wall on the other side*.” Generating (indefinite) referring expressions to objects that do not exist yet is fundamentally different from generating the usual definite referring expressions, because the new object must be distinguished from an infinite set of possible alternatives. We formalize this problem and present an algorithm for generating such expressions, in the context of generating building instructions within the Minecraft video game.

We model the production of quantified referring expressions (QREs) that identify collections of visual items. To address this task, we propose a method of perceptual cost pruning, which consists of two steps: (1) determine what subset of quantity information can be perceived given a time limit t, and (2) apply a preference order based REG algorithm (e.g., the Incremental Algorithm) to this reduced set of information. We demonstrate that this method successfully improves the human-likeness of the IA in the QRE generation task and successfully models human-generated language in most cases.

In recent years, Bayesian models of referring expression generation have gained prominence in order to produce situationally more adequate referring expressions. Basically, these models enable the integration of different parameters into the decision process for using a specific referring expression like the cardinality of the object set, the configuration and complexity of the visual field, and the discriminatory power of available attributes that need to be combined with visual salience and personal preference. This paper describes and discusses the results of an empirical study on the production of referring expressions in visual fields with different object configurations of varying complexity and different contextual premises for using a referring expression. The visual fields are set up using data from the TUNA experiment with plain random or pragmatically enriched configurations which allow for target inference. Different categories of the situational contexts, in which the referring expressions are produced, provide different degrees of cooperativeness, so that generation quality and its relations to contextual user intention can be observed. The results of the study suggest that Bayesian approaches must integrate individual generation preference and the cooperativeness of the situational task in order to model the broad variance between speakers more adequately.

Text attribute transfer is modifying certain linguistic attributes (e.g. sentiment, style, author-ship, etc.) of a sentence and transforming them from one type to another. In this paper, we aim to analyze and interpret what is changed during the transfer process. We start from the observation that in many existing models and datasets, certain words within a sentence play important roles in determining the sentence attribute class. These words are referred as the Pivot Words. Based on these pivot words, we propose a lexical analysis framework, the Pivot Analysis, to quantitatively analyze the effects of these words in text attribute classification and transfer. We apply this framework to existing datasets and models and show that: (1) the pivot words are strong features for the classification of sentence attributes; (2) to change the attribute of a sentence, many datasets only requires to change certain pivot words; (3) consequently, many transfer models only perform the lexical-level modification,while leaving higher-level sentence structures unchanged. Our work provides an in-depth understanding of linguistic attribute transfer and further identifies the future requirements and challenges of this task

Between 80% and 90% of all Chinese words have long and short form such as 老虎/虎 (lao-hu/hu , tiger) (Duanmu:2013). Consequently, the choice between long and short forms is a key problem for lexical choice across NLP and NLG. Following an earlier work on abbreviations in English (Mahowald et al, 2013), we bring a probabilistic perspective to these questions, using both a behavioral and a corpus-based approach. We hypothesized that there is a higher probability of choosing short form in supportive context than in neutral context in Mandarin. Consistent with our prediction, our findings revealed that predictability of contexts makes effect on speakers’ long and short form choice.

The writing process consists of several stages such as drafting, revising, editing, and proofreading. Studies on writing assistance, such as grammatical error correction (GEC), have mainly focused on sentence editing and proofreading, where surface-level issues such as typographical errors, spelling errors, or grammatical errors should be corrected. We broaden this focus to include the earlier revising stage, where sentences require adjustment to the information included or major rewriting and propose Sentence-level Revision (SentRev) as a new writing assistance task. Well-performing systems in this task can help inexperienced authors by producing fluent, complete sentences given their rough, incomplete drafts. We build a new freely available crowdsourced evaluation dataset consisting of incomplete sentences authored by non-native writers paired with their final versions extracted from published academic papers for developing and evaluating SentRev models. We also establish baseline performance on SentRev using our newly built evaluation dataset.

Synthesis approaches in computational argumentation so far are restricted to generating claim-like argument units or short summaries of debates. Ultimately, however, we expect computers to generate whole new arguments for a given stance towards some topic, backing up claims following argumentative and rhetorical considerations. In this paper, we approach such an argumentation synthesis as a language modeling task. In our language model, argumentative discourse units are the “words”, and arguments represent the “sentences”. Given a pool of units for any unseen topic-stance pair, the model selects a set of unit types according to a basic rhetorical strategy (logos vs. pathos), arranges the structure of the types based on the units’ argumentative roles, and finally “phrases” an argument by instantiating the structure with semantically coherent units from the pool. Our evaluation suggests that the model can, to some extent, mimic the human synthesis of strategy-specific arguments.

Encoder-decoder based neural architectures serve as the basis of state-of-the-art approaches in end-to-end open domain dialog systems. Since most of such systems are trained with a maximum likelihood (MLE) objective they suffer from issues such as lack of generalizability and the generic response problem, i.e., a system response that can be an answer to a large number of user utterances, e.g., “Maybe, I don’t know.” Having explicit feedback on the relevance and interestingness of a system response at each turn can be a useful signal for mitigating such issues and improving system quality by selecting responses from different approaches. Towards this goal, we present a system that evaluates chatbot responses at each dialog turn for coherence and engagement. Our system provides explicit turn-level dialog quality feedback, which we show to be highly correlated with human evaluation. To show that incorporating this feedback in the neural response generation models improves dialog quality, we present two different and complementary mechanisms to incorporate explicit feedback into a neural response generation model: reranking and direct modification of the loss function during training. Our studies show that a response generation model that incorporates these combined feedback mechanisms produce more engaging and coherent responses in an open-domain spoken dialog setting, significantly improving the response quality using both automatic and human evaluation.

We investigate the impact of search strategies in neural dialogue modeling. We first compare two standard search algorithms, greedy and beam search, as well as our newly proposed iterative beam search which produces a more diverse set of candidate responses. We evaluate these strategies in realistic full conversations with humans and propose a model-based Bayesian calibration to address annotator bias. These conversations are analyzed using two automatic metrics: log-probabilities assigned by the model and utterance diversity. Our experiments reveal that better search algorithms lead to higher rated conversations. However, finding the optimal selection mechanism to choose from a more diverse set of candidates is still an open question.

To overcome the limitations of automated metrics (e.g. BLEU, METEOR) for evaluating dialogue systems, researchers typically use human judgments to provide convergent evidence. While it has been demonstrated that human judgments can suffer from the inconsistency of ratings, extant research has also found that the design of the evaluation task affects the consistency and quality of human judgments. We conduct a between-subjects study to understand the impact of four experiment conditions on human ratings of dialogue system output. In addition to discrete and continuous scale ratings, we also experiment with a novel application of Best-Worst scaling to dialogue evaluation. Through our systematic study with 40 crowdsourced workers in each task, we find that using continuous scales achieves more consistent ratings than Likert scale or ranking-based experiment design. Additionally, we find that factors such as time taken to complete the task and no prior experience of participating in similar studies of rating dialogue system output positively impact consistency and agreement amongst raters.

Generating fluent natural language responses from structured semantic representations is a critical step in task-oriented conversational systems. Sequence-to-sequence models on flat meaning representations (MR) have been dominant in this task, for example in the E2E NLG Challenge. Previous work has shown that a tree-structured MR can improve the model for better discourse-level structuring and sentence-level planning. In this work, we propose a tree-to-sequence model that uses a tree-LSTM encoder to leverage the tree structures in the input MR, and further enhance the decoding by a structure-enhanced attention mechanism. In addition, we explore combining these enhancements with constrained decoding to improve semantic correctness. Our experiments not only show significant improvements over standard seq2seq baselines, but also is more data-efficient and generalizes better to hard scenarios.

Multiple headlines of a newspaper article have an important role to express the content of the article accurately and concisely. A headline depends on the content and intent of their article. While a single headline expresses the whole corresponding article, each of multiple headlines expresses different information individually. We suggest automatic generation method of such a diverse multiple headlines in a newspaper. Our generation method is based on the Pointer-Generator Network, using page metadata on a newspaper which can change headline generation behavior. This page metadata includes headline location, headline size, article page number, etc. In a previous related work, ensemble of three different generation models was performed to obtain a single headline, where each generation model generates a single headline candidate. In contrast, we use a single model to generate multiple headlines. We conducted automatic evaluations for generated headlines. The results show that our method improved ROUGE-1 score by 4.32 points higher than baseline. These results suggest that our model using page metadata can generate various multiple headlines for an article In better performance.

Question Generation (QG) is the task of generating questions from a given passage. One of the key requirements of QG is to generate a question such that it results in a target answer. Previous works used a target answer to obtain a desired question. However, we also want to specify how to ask questions and improve the quality of generated questions. In this study, we explore the use of interrogative phrases as additional sources to control QG. By providing interrogative phrases, we expect that QG can generate a more reliable sequence of words subsequent to an interrogative phrase. We present a baseline sequence-to-sequence model with the attention, copy, and coverage mechanisms, and show that the simple baseline achieves state-of-the-art performance. The experiments demonstrate that interrogative phrases contribute to improving the performance of QG. In addition, we report the superiority of using interrogative phrases in human evaluation. Finally, we show that a question answering system can provide target answers more correctly when the questions are generated with interrogative phrases.

Surface realisation (SR) consists in generating a text from a meaning representations (MR). In this paper, we introduce a new parallel dataset of deep meaning representations (MR) and French sentences and we present a novel method for MR-to-text generation which seeks to generalise by abstracting away from lexical content. Most current work on natural language generation focuses on generating text that matches a reference using BLEU as evaluation criteria. In this paper, we additionally consider the model’s ability to reintroduce the function words that are absent from the deep input meaning representations. We show that our approach increases both BLEU score and the scores used to assess function words generation.

We compiled a new sentence splitting corpus that is composed of 203K pairs of aligned complex source and simplified target sentences. Contrary to previously proposed text simplification corpora, which contain only a small number of split examples, we present a dataset where each input sentence is broken down into a set of minimal propositions, i.e. a sequence of sound, self-contained utterances with each of them presenting a minimal semantic unit that cannot be further decomposed into meaningful propositions. This corpus is useful for developing sentence splitting approaches that learn how to transform sentences with a complex linguistic structure into a fine-grained representation of short sentences that present a simple and more regular structure which is easier to process for downstream applications and thus facilitates and improves their performance.

A prominent strand of work in formal semantics investigates the ways in which human languages quantify over the elements of a set, as when we say “All A are B”, “All except two A are B”, “Only a few of the A are B” and so on. Our aim is to build Natural Language Generation algorithms that mimic humans’ use of quantified expressions. To inform these algorithms, we conducted on a series of elicitation experiments in which human speakers were asked to perform a linguistic task that invites the use of quantified expressions. We discuss how these experiments were conducted and what corpora they gave rise to. We conduct an informal analysis of the corpora, and offer an initial assessment of the challenges that these corpora pose for Natural Language Generation. The dataset is available at: https://github.com/a-quei/qtuna.

Keyphrase generation is the task of predicting a set of lexical units that conveys the main content of a source text. Existing datasets for keyphrase generation are only readily available for the scholarly domain and include non-expert annotations. In this paper we present KPTimes, a large-scale dataset of news texts paired with editor-curated keyphrases. Exploring the dataset, we show how editors tag documents, and how their annotations differ from those found in existing datasets. We also train and evaluate state-of-the-art neural keyphrase generation models on KPTimes to gain insights on how well they perform on the news domain. The dataset is available online at https://github.com/ygorg/KPTimes.

A lot of recent work in Language & Vision has looked at generating descriptions or referring expressions for objects in scenes of real-world images, though focusing mostly on relatively simple language like object names, color and location attributes (e.g., brown chair on the left). This paper presents work on Draw-and-Tell, a dataset of detailed descriptions for common objects in images where annotators have produced fine-grained attribute-centric expressions distinguishing a target object from a range of similar objects. Additionally, the dataset comes with hand-drawn sketches for each object. As Draw-and-Tell is medium-sized and contains a rich vocabulary, it constitutes an interesting challenge for CNN-LSTM architectures used in state-of-the-art image captioning models. We explore whether the additional modality given through sketches can help such a model to learn to accurately ground detailed language referring expressions to object shapes. Our results are encouraging.

Data-to-text generation aims to generate descriptions given a structured input data (i.e., a table with multiple records). Existing neural methods for encoding input data can be divided into two categories: a) pooling based encoders which ignore dependencies between input records or b) recurrent encoders which model only sequential dependencies between input records. In our investigation, although the recurrent encoder generally outperforms the pooling based encoder by learning the sequential dependencies, it is sensitive to the order of the input records (i.e., performance decreases when injecting the random shuffling noise over input data). To overcome this problem, we propose to adopt the self-attention mechanism to learn dependencies between arbitrary input records. Experimental results show the proposed method achieves comparable results and remains stable under random shuffling over input data.

Leveraging the visual modality effectively for Neural Machine Translation (NMT) remains an open problem in computational linguistics. Recently, Caglayan et al. posit that the observed gains are limited mainly due to the very simple, short, repetitive sentences of the Multi30k dataset (the only multimodal MT dataset available at the time), which renders the source text sufficient for context. In this work, we further investigate this hypothesis on a new large scale multimodal Machine Translation (MMT) dataset, How2, which has 1.57 times longer mean sentence length than Multi30k and no repetition. We propose and evaluate three novel fusion techniques, each of which is designed to ensure the utilization of visual context at different stages of the Sequence-to-Sequence transduction pipeline, even under full linguistic context. However, we still obtain only marginal gains under full linguistic context and posit that visual embeddings extracted from deep vision models (ResNet for Multi30k, ResNext for How2) do not lend themselves to increasing the discriminativeness between the vocabulary elements at token level prediction in NMT. We demonstrate this qualitatively by analyzing attention distribution and quantitatively through Principal Component Analysis, arriving at the conclusion that it is the quality of the visual embeddings rather than the length of sentences, which need to be improved in existing MMT datasets.

We present a dataset consisting of what we call image description sequences, which are multi-sentence descriptions of the contents of an image. These descriptions were collected in a pseudo-interactive setting, where the describer was told to describe the given image to a listener who needs to identify the image within a set of images, and who successively asks for more information. As we show, this setup produced nicely structured data that, we think, will be useful for learning models capable of planning and realising such description discourses.

Automatic natural language generation systems need to use the contextually-appropriate verbs when describing different kinds of facts or events, which has triggered research interest on verb selection for data-to-text generation. In this paper, we discuss a few limitations of the current task settings and the evaluation metrics. We also provide two simple, efficient, interpretable baseline approaches for statistical selection of trend verbs, which give a strong performance on both previously used evaluation metrics and our new evaluation.

The uptake of deep learning in natural language generation (NLG) led to the release of both small and relatively large parallel corpora for training neural models. The existing data-to-text datasets are, however, aimed at task-oriented dialogue systems, and often thus limited in diversity and versatility. They are typically crowdsourced, with much of the noise left in them. Moreover, current neural NLG models do not take full advantage of large training data, and due to their strong generalizing properties produce sentences that look template-like regardless. We therefore present a new corpus of 7K samples, which (1) is clean despite being crowdsourced, (2) has utterances of 9 generalizable and conversational dialogue act types, making it more suitable for open-domain dialogue systems, and (3) explores the domain of video games, which is new to dialogue systems despite having excellent potential for supporting rich conversations.

In this study, we investigate the employment of the pre-trained BERT language model to tackle question generation tasks. We introduce two neural architectures built on top of BERT for question generation tasks. The first one is a straightforward BERT employment, which reveals the defects of directly using BERT for text generation. And, the second one remedies the first one by restructuring the BERT employment into a sequential manner for taking information from previous decoded results. Our models are trained and evaluated on the question-answering dataset SQuAD. Experiment results show that our best model yields state-of-the-art performance which advances the BLEU4 score of existing best models from 16.85 to 18.91.

Natural Language Inference (NLI) is the task of determining the semantic relationship between a premise and a hypothesis. In this paper, we focus on the generation of hypotheses from premises in a multimodal setting, to generate a sentence (hypothesis) given an image and/or its description (premise) as the input. The main goals of this paper are (a) to investigate whether it is reasonable to frame NLI as a generation task; and (b) to consider the degree to which grounding textual premises in visual information is beneficial to generation. We compare different neural architectures, showing through automatic and human evaluation that entailments can indeed be generated successfully. We also show that multimodal models outperform unimodal models in this task, albeit marginally

Machine-translated text plays a crucial role in the communication of people using different languages. However, adversaries can use such text for malicious purposes such as plagiarism and fake review. The existing methods detected a machine-translated text only using the text’s intrinsic content, but they are unsuitable for classifying the machine-translated and human-written texts with the same meanings. We have proposed a method to extract features used to distinguish machine/human text based on the similarity between the intrinsic text and its back-translation. The evaluation of detecting translated sentences with French shows that our method achieves 75.0% of both accuracy and F-score. It outperforms the existing methods whose the best accuracy is 62.8% and the F-score is 62.7%. The proposed method even detects more efficiently the back-translated text with 83.4% of accuracy, which is higher than 66.7% of the best previous accuracy. We also achieve similar results not only with F-score but also with similar experiments related to Japanese. Moreover, we prove that our detector can recognize both machine-translated and machine-back-translated texts without the language information which is used to generate these machine texts. It demonstrates the persistence of our method in various applications in both low- and rich-resource languages.

Building a controllable neural conversation model (NCM) is an important task. In this paper, we focus on controlling the responses of NCMs by using dialogue act labels of responses as conditions. We introduce an adversarial learning framework for the task of generating conditional responses with a new objective to a discriminator, which explicitly distinguishes sentences by using labels. This change strongly encourages the generation of label-conditioned sentences. We compared the proposed method with some existing methods for generating conditional responses. The experimental results show that our proposed method has higher controllability for dialogue acts even though it has higher or comparable naturalness to existing methods.

Despite the success of style transfer in image processing, it has seen limited progress in natural language generation. Part of the problem is that content is not as easily decoupled from style in the text domain. Curiously, in the field of stylometry, content does not figure prominently in practical methods of discriminating stylistic elements, such as authorship and genre. Rather, syntax and function words are the most salient features. Drawing on this work, we model style as a suite of low-level linguistic controls, such as frequency of pronouns, prepositions, and subordinate clause constructions. We train a neural encoder-decoder model to reconstruct reference sentences given only content words and the setting of the controls. We perform style transfer by keeping the content words fixed while adjusting the controls to be indicative of another style. In experiments, we show that the model reliably responds to the linguistic controls and perform both automatic and manual evaluations on style transfer. We find we can fool a style classifier 84% of the time, and that our model produces highly diverse and stylistically distinctive outputs. This work introduces a formal, extendable model of style that can add control to any neural text generation system.

Neural Machine Translation (NMT) models tend to achieve the best performances when larger sets of parallel sentences are provided for training. For this reason, augmenting the training set with artificially-generated sentence pair can boost the performance. Nonetheless, the performance can also be improved with a small number of sentences if they are in the same domain as the test set. Accordingly, we want to explore the use of artificially-generated sentence along with data-selection algorithms to improve NMT models trained solely with authentic data. In this work, we show how artificially-generated sentences can be more beneficial than authentic pairs and what are their advantages when used in combination with data-selection algorithms.

Using data-driven models for solving text summarization or similar tasks has become very common in the last years. Yet most of the studies report basic accuracy scores only, and nothing is known about the ability of the proposed models to improve when trained on more data. In this paper, we define and propose three data efficiency metrics: data score efficiency, data time deficiency and overall data efficiency. We also propose a simple scheme that uses those metrics and apply it for a more comprehensive evaluation of popular methods on text summarization and title generation tasks. For the latter task, we process and release a huge collection of 35 million abstract-title pairs from scientific articles. Our results reveal that among the tested models, the Transformer is the most efficient on both tasks.

Roughly 30% of congressional staffers in the United States report spending a “great deal” of time writing responses to constituent letters. Letters often solicit an update on the status of legislation and a description of a congressman’s vote record or vote intention — structurable data that can be leveraged by a natural language generation (NLG) system to create a coherent letter response. This paper describes how PoliScribe, a pipeline-architectured NLG platform, constructs personalized responses to constituents inquiring about legislation. Emphasis will be placed on adapting NLG methodologies to the political domain, which entails special attention to affect, discursive variety, and rhetorical strategies that align a speaker with their interlocutor, even in cases of policy disagreement.

We present Margin Call, a web-based text viewer that automatically generates short summaries for each paragraph of the text and displays the summaries in the margin of the text next to the corresponding paragraph. On the back-end, the summarizer first identifies the most important sentence for each paragraph in the text file uploaded by the user. The selected sentence is then automatically compressed to produce the short summary. The resulting summary is a few words long. The displayed summaries can help the user understand and retrieve information faster from the text, while increasing the retention of information.

Biased decisions made by automatic systems have led to growing concerns in research communities. Recent work from the NLP community focuses on building systems that make fair decisions based on text. Instead of relying on unknown decision systems or human decision-makers, we argue that a better way to protect data providers is to remove the trails of sensitive information before publishing the data. In light of this, we propose a new privacy-aware text rewriting task and explore two privacy-aware back-translation methods for the task, based on adversarial training and approximate fairness risk. Our extensive experiments on three real-world datasets with varying demographical attributes show that our methods are effective in obfuscating sensitive attributes. We have also observed that the fairness risk method retains better semantics and fluency, while the adversarial training method tends to leak less sensitive information.

A lexical simplification (LS) system substitutes difficult words in a text with simpler ones to make it easier for the user to understand. In the typical LS pipeline, the Substitution Ranking step determines the best substitution out of a set of candidates. Most current systems do not consider the user’s vocabulary proficiency, and always aim for the simplest candidate. This approach may overlook less-simple candidates that the user can understand, and that are semantically closer to the original word. We propose a personalized approach for Substitution Ranking to identify the candidate that is the closest synonym and is non-complex for the user. In experiments on learners of English at different proficiency levels, we show that this approach enhances the semantic faithfulness of the output, at the cost of a relatively small increase in the number of complex words.

End-to-end neural approaches have achieved state-of-the-art performance in many natural language processing (NLP) tasks. Yet, they often lack transparency of the underlying decision-making process, hindering error analysis and certain model improvements. In this work, we revisit the binary linearization approach to surface realization, which exhibits more interpretable behavior, but was falling short in terms of prediction accuracy. We show how enriching the training data to better capture word order constraints almost doubles the performance of the system. We further demonstrate that encoding both local and global prediction contexts yields another considerable performance boost. With the proposed modifications, the system which ranked low in the latest shared task on multilingual surface realization now achieves best results in five out of ten languages, while being on par with the state-of-the-art approaches in others.

We present a dependency tree linearization model with two novel components: (1) a tree-structured encoder based on bidirectional Tree-LSTM that propagates information first bottom-up then top-down, which allows each token to access information from the entire tree; and (2) a linguistically motivated head-first decoder that emphasizes the central role of the head and linearizes the subtree by incrementally attaching the dependents on both sides of the head. With the new encoder and decoder, we reach state-of-the-art performance on the Surface Realization Shared Task 2018 dataset, outperforming not only the shared tasks participants, but also previous state-of-the-art systems (Bohnet et al., 2011; Puduppully et al., 2016). Furthermore, we analyze the power of the tree-structured encoder with a probing task and show that it is able to recognize the topological relation between any pair of tokens in a tree.

We present a creative poem generator for the morphologically rich Finnish language. Our method falls into the master-apprentice paradigm, where a computationally creative genetic algorithm teaches a BRNN model to generate poetry. We model several parts of poetic aesthetics in the fitness function of the genetic algorithm, such as sonic features, semantic coherence, imagery and metaphor. Furthermore, we justify the creativity of our method based on the FACE theory on computational creativity and take additional care in evaluating our system by automatic metrics for concepts together with human evaluation for aesthetics, framing and expressions.

This paper addresses Hip-Hop lyric generation with conditional Neural Language Models. We develop a simple yet effective mechanism to extract and apply conditional templates from text snippets, and show—on the basis of a large-scale crowd-sourced manual evaluation—that these templates significantly improve the quality and realism of the generated snippets. Importantly, the proposed approach enables end-to-end training, targeting formal properties of text such as rhythm and rhyme, which are central characteristics of rap texts. Additionally, we explore how generating text at different scales (e.g. character-level or word-level) affects the quality of the output. We find that a hybrid form—a hierarchical model that aims to integrate Language Modeling at both word and character-level scales—yields significant improvements in text quality, yet surprisingly, cannot exploit conditional templates to their fullest extent. Our findings highlight that text generation models based on Recurrent Neural Networks (RNN) are sensitive to the modeling scale and call for further research on the observed differences in effectiveness of the conditioning mechanism at different scales.

Data-to-text generation models face challenges in ensuring data fidelity by referring to the correct input source. To inspire studies in this area, Wiseman et al. (2017) introduced the RotoWire corpus on generating NBA game summaries from the box- and line-score tables. However, limited attempts have been made in this direction and the challenges remain. We observe a prominent bottleneck in the corpus where only about 60% of the summary contents can be grounded to the boxscore records. Such information deficiency tends to misguide a conditioned language model to produce unconditioned random facts and thus leads to factual hallucinations. In this work, we restore the information balance and revamp this task to focus on fact-grounded data-to-text generation. We introduce a purified and larger-scale dataset, RotoWire-FG (Fact-Grounding), with 50% more data from the year 2017-19 and enriched input tables, and hope to attract research focuses in this direction. Moreover, we achieve improved data fidelity over the state-of-the-art models by integrating a new form of table reconstruction as an auxiliary task to boost the generation quality.

We propose a data-to-document generator that can easily control the contents of output texts based on a neural language model. Conventional data-to-text model is useful when a reader seeks a global summary of data because it has only to describe an important part that has been extracted beforehand. However, because depending on users, it differs what they are interested in, so it is necessary to develop a method to generate various summaries according to users’ interests. We develop a model to generate various summaries and to control their contents by providing the explicit targets for a reference to the model as controllable factors. In the experiments, we used five-minute or one-hour charts of 9 indicators (e.g., Nikkei225), as time-series data, and daily summaries of Nikkei Quick News as textual data. We conducted comparative experiments using two pieces of information: human-designed topic labels indicating the contents of a sentence and automatically extracted keywords as the referential information for generation.

Browsing news articles on multiple devices is now possible. The lengths of news article headlines have precise upper bounds, dictated by the size of the display of the relevant device or interface. Therefore, controlling the length of headlines is essential when applying the task of headline generation to news production. However, because there is no corpus of headlines of multiple lengths for a given article, previous research on controlling output length in headline generation has not discussed whether the system outputs could be adequately evaluated without multiple references of different lengths. In this paper, we introduce two corpora, which are Japanese News Corpus (JNC) and JApanese MUlti-Length Headline Corpus (JAMUL), to confirm the validity of previous evaluation settings. The JNC provides common supervision data for headline generation. The JAMUL is a large-scale evaluation dataset for headlines of three different lengths composed by professional editors. We report new findings on these corpora; for example, although the longest length reference summary can appropriately evaluate the existing methods controlling output length, this evaluation setting has several problems.

Inter-Annotator Agreement (IAA) is used as a means of assessing the quality of NLG evaluation data, in particular, its reliability. According to existing scales of IAA interpretation – see, for example, Lommel et al. (2014), Liu et al. (2016), Sedoc et al. (2018) and Amidei et al. (2018a) – most data collected for NLG evaluation fail the reliability test. We confirmed this trend by analysing papers published over the last 10 years in NLG-specific conferences (in total 135 papers that included some sort of human evaluation study). Following Sampson and Babarczy (2008), Lommel et al. (2014), Joshi et al. (2016) and Amidei et al. (2018b), such phenomena can be explained in terms of irreducible human language variability. Using three case studies, we show the limits of considering IAA as the only criterion for checking evaluation reliability. Given human language variability, we propose that for human evaluation of NLG, correlation coefficients and agreement coefficients should be used together to obtain a better assessment of the evaluation data reliability. This is illustrated using the three case studies.

Currently, there is little agreement as to how Natural Language Generation (NLG) systems should be evaluated. While there is some agreement regarding automatic metrics, there is a high degree of variation in the way that human evaluation is carried out. This paper provides an overview of how human evaluation is currently conducted, and presents a set of best practices, grounded in the literature. With this paper, we hope to contribute to the quality and consistency of human evaluations in NLG.

We present a recurrent neural network based system for automatic quality estimation of natural language generation (NLG) outputs, which jointly learns to assign numerical ratings to individual outputs and to provide pairwise rankings of two different outputs. The latter is trained using pairwise hinge loss over scores from two copies of the rating network. We use learning to rank and synthetic data to improve the quality of ratings assigned by our system: We synthesise training pairs of distorted system outputs and train the system to rank the less distorted one higher. This leads to a 12% increase in correlation with human ratings over the previous benchmark. We also establish the state of the art on the dataset of relative rankings from the E2E NLG Challenge (Dusek et al., 2019), where synthetic data lead to a 4% accuracy increase over the base model.

We follow the step-by-step approach to neural data-to-text generation proposed by Moryossef et al (2019), in which the generation process is divided into a text planning stage followed by a plan realization stage. We suggest four extensions to that framework: (1) we introduce a trainable neural planning component that can generate effective plans several orders of magnitude faster than the original planner; (2) we incorporate typing hints that improve the model’s ability to deal with unseen relations and entities; (3) we introduce a verification-by-reranking stage that substantially improves the faithfulness of the resulting texts; (4) we incorporate a simple but effective referring expression generation module. These extensions result in a generation process that is faster, more fluent, and more accurate.

Creativity is an essential element of human nature used for many activities, such as telling a story. Based on human creativity, researchers have attempted to teach a computer to generate stories automatically or support this creative process. In this study, we undertake the task of story ending generation. This is a relatively new task, in which the last sentence of a given incomplete story is automatically generated. This is challenging because, in order to predict an appropriate ending, the generation method should comprehend the context of events. Despite the importance of this task, no clear evaluation metric has been established thus far; hence, it has remained an open problem. Therefore, we study the various elements involved in evaluating an automatic method for generating story endings. First, we introduce a baseline hierarchical sequence-to-sequence method for story ending generation. Then, we conduct a pairwise comparison against human-written endings, in which annotators choose the preferable ending. In addition to a quantitative evaluation, we conduct a qualitative evaluation by asking annotators to specify the reason for their choice. From the collected reasons, we discuss what elements the evaluation should focus on, to thereby propose effective metrics for the task.

This paper describes the implementation of the Hotel Scribe system. A commercial Natural Language Generation (NLG) system which generates descriptions of hotels from accommodation metadata with a high level of content and linguistic variation in English. It has been deployed live by *Anonymised Company Name* for the purpose of improving coverage of accommodation descriptions and for Search Engine Optimisation (SEO). In this paper, we describe the motivation for building this system, the challenges faced when dealing with limited metadata, and the implementation used to generate the highly variate accommodation descriptions. Additionally, we evaluate the uniqueness of the texts generated by our system against comparable human written accommodation description texts.

Rating and Likert scales are widely used in evaluation experiments to measure the quality of Natural Language Generation (NLG) systems. We review the use of rating and Likert scales for NLG evaluation tasks published in NLG specialized conferences over the last ten years (135 papers in total). Our analysis brings to light a number of deviations from good practice in their use. We conclude with some recommendations about the use of such scales. Our aim is to encourage the appropriate use of evaluation methodologies in the NLG community.

Task effects in NLG corpus elicitation recently started to receive more attention, but are usually not modeled statistically. We present a controlled replication of the study by Van Miltenburg et al. (2018b), contrasting spoken with written descriptions. We collected additional written Dutch descriptions to supplement the spoken data from the DIDEC corpus, and analyzed the descriptions using mixed effects modeling to account for variation between participants and items. Our results show that the effects of modality largely disappear in a controlled setting.

Multimedia procedural texts, such as instructions and manuals with pictures, support people to share how-to knowledge. In this paper, we propose a method for generating a procedural text given a photo sequence allowing users to obtain a multimedia procedural text. We propose a single embedding space both for image and text enabling to interconnect them and to select appropriate words to describe a photo. We implemented our method and tested it on cooking instructions, i.e., recipes. Various experimental results showed that our method outperforms standard baselines.

SimpleNLG is a popular open source surface realiser for the English language. For German, however, the availability of open source and non-domain specific realisers is sparse, partly due to the complexity of the German language. In this paper, we present SimpleNLG-DE, an adaption of SimpleNLG to German. We discuss which parts of the German language have been implemented and how we evaluated our implementation using the TIGER Corpus and newly created data-sets.

Neural natural language generation (NNLG) systems are known for their pathological outputs, i.e. generating text which is unrelated to the input specification. In this paper, we show the impact of semantic noise on state-of-the-art NNLG models which implement different semantic control mechanisms. We find that cleaned data can improve semantic correctness by up to 97%, while maintaining fluency. We also find that the most common error is omitting information, rather than hallucination.

Learned dynamic weighting of the conditioning signal (attention) has been shown to improve neural language generation in a variety of settings. The weights applied when generating a particular output sequence have also been viewed as providing a potentially explanatory insight in the internal workings of the generator. In this paper, we reverse the direction of this connection and ask whether through the control of the attention of the model we can control its output. Specifically, we take a standard neural image captioning model that uses attention, and fix the attention to predetermined areas in the image. We evaluate whether the resulting output is more likely to mention the class of the object in that area than the normally generated caption. We introduce three effective methods to control the attention and find that these are producing expected results in up to 27.43% of the cases.

We present “AutoJudge”, an automated evaluation method for conversational dialogue systems. The method works by first generating dialogues based on self-talk, i.e. dialogue systems talking to itself. Then, it uses human ratings on these dialogues to train an automated judgement model. Our experiments show that AutoJudge correlates well with the human ratings and can be used to automatically evaluate dialogue systems, even in deployed systems. In a second part, we attempt to apply AutoJudge to improve existing systems. This works well for re-ranking a set of candidate utterances. However, our experiments show that AutoJudge cannot be applied as reward for reinforcement learning, although the metric can distinguish good from bad dialogues. We discuss potential reasons, but state here already that this is still an open question for further research.

In this work, we examine whether it is possible to achieve the state of the art performance in paraphrase generation with reduced vocabulary. Our approach consists of building a convolution to sequence model (Conv2Seq) partially guided by the reinforcement learning, and training it on the subword representation of the input. The experiment on the Quora dataset, which contains over 140,000 pairs of sentences and corresponding paraphrases, found that with less than 1,000 token types, we were able to achieve performance which exceeded that of the current state of the art.

In this paper, we present a novel data-to-text system for cancer patients, providing information on quality of life implications after treatment, which can be embedded in the context of shared decision making. Currently, information on quality of life implications is often not discussed, partly because (until recently) data has been lacking. In our work, we rely on a newly developed prediction model, which assigns patients to scenarios. Furthermore, we use data-to-text techniques to explain these scenario-based predictions in personalized and understandable language. We highlight the possibilities of NLG for personalization, discuss ethical implications and also present the outcomes of a first evaluation with clinicians.

Current approaches to Natural Language Generation (NLG) for dialog mainly focus on domain-specific, task-oriented applications (e.g. restaurant booking) using limited ontologies (up to 20 slot types), usually without considering the previous conversation context. Furthermore, these approaches require large amounts of data for each domain, and do not benefit from examples that may be available for other domains. This work explores the feasibility of applying statistical NLG to scenarios requiring larger ontologies, such as multi-domain dialog applications or open-domain question answering (QA) based on knowledge graphs. We model NLG through an Encoder-Decoder framework using a large dataset of interactions between real-world users and a conversational agent for open-domain QA. First, we investigate the impact of increasing the number of slot types on the generation quality and experiment with different partitions of the QA data with progressively larger ontologies (up to 369 slot types). Second, we perform multi-task learning experiments between open-domain QA and task-oriented dialog, and benchmark our model on a popular NLG dataset. Moreover, we experiment with using the conversational context as an additional input to improve response generation quality. Our experiments show the feasibility of learning statistical NLG models for open-domain QA with larger ontologies.

People with sight impairments can access to a mathematical expression by using its LaTeX source. However, this mechanisms have several drawbacks: (1) it assumes the knowledge of the LaTeX, (2) it is slow, since LaTeX is verbose and (3) it is error-prone since LATEX is a typographical language. In this paper we study the design of a natural language generation system for producing a mathematical sentence, i.e. a natural language sentence expressing the semantics of a mathematical expression. Moreover, we describe the main results of a first human based evaluation experiment of the system for Italian language.

Statistical generators increasingly dominate the research in NLG. However, grammar-based generators that are grounded in a solid linguistic framework remain very competitive, especially for generation from deep knowledge structures. Furthermore, if built modularly, they can be ported to other genres and languages with a limited amount of work, without the need of the annotation of a considerable amount of training data. One of these generators is FORGe, which is based on the Meaning-Text Model. In the recent WebNLG challenge (the first comprehensive task addressing the mapping of RDF triples to text) FORGe ranked first with respect to the overall quality in human evaluation. We extend the coverage of FORGE’s open source grammatical and lexical resources for English, so as to further improve the English texts, and port them to Spanish, to achieve a comparable quality. This confirms that, as already observed in the case of SimpleNLG, a robust universal grammar-driven framework and a systematic organization of the linguistic resources can be an adequate choice for NLG applications.

We present an approach to generating natural language justifications of decisions derived from norm-based reasoning. Assuming an agent which maximally satisfies a set of rules specified in an object-oriented temporal logic, the user can ask factual questions (about the agent’s rules, actions, and the extent to which the agent violated the rules) as well as “why” questions that require the agent comparing actual behavior to counterfactual trajectories with respect to these rules. To produce natural-sounding explanations, we focus on the subproblem of producing natural language clauses from statements in a fragment of temporal logic, and then describe how to embed these clauses into explanatory sentences. We use a human judgment evaluation on a testbed task to compare our approach to variants in terms of intelligibility, mental model and perceived trust.

A math word problem is a narrative with a specific topic that provides clues to the correct equation with numerical quantities and variables therein. In this paper, we focus on the task of generating math word problems. Previous works are mainly template-based with pre-defined rules. We propose a novel neural network model to generate math word problems from the given equations and topics. First, we design a fusion mechanism to incorporate the information of both equations and topics. Second, an entity-enforced loss is introduced to ensure the relevance between the generated math problem and the equation. Automatic evaluation results show that the proposed model significantly outperforms the baseline models. In human evaluations, the math word problems generated by our model are rated as being more relevant (in terms of solvability of the given equations and relevance to topics) and natural (i.e., grammaticality, fluency) than the baseline models.

We introduce DisSim, a discourse-aware sentence splitting framework for English and German whose goal is to transform syntactically complex sentences into an intermediate representation that presents a simple and more regular structure which is easier to process for downstream semantic applications. For this purpose, we turn input sentences into a two-layered semantic hierarchy in the form of core facts and accompanying contexts, while identifying the rhetorical relations that hold between them. In that way, we preserve the coherence structure of the input and, hence, its interpretability for downstream tasks.

This study presents a voice assistant system to support cooking by utilizing smart speakers in Japan. This system not only speaks the procedures written in recipes point by point but also answers the common questions from users for the specified recipes. The system applies machine comprehension techniques to millions of recipes for answering the common questions in cooking such as “人参はどうしたらよいですか (How should I cook carrots?)”. Furthermore, numerous machine-learning techniques are applied to generate better responses to users.

This paper describes our submission to the TL;DR challenge. Neural abstractive summarization models have been successful in generating fluent and consistent summaries with advancements like the copy (Pointer-generator) and coverage mechanisms. However, these models suffer from their extractive nature as they learn to copy words from the source text. In this paper, we propose a novel abstractive model based on Variational Autoencoder (VAE) to address this issue. We also propose a Unified Summarization Framework for the generation of summaries. Our model eliminates non-critical information at a sentence-level with an extractive summarization module and generates the summary word by word using an abstractive summarization module. To implement our framework, we combine submodules with state-of-the-art techniques including Pointer-Generator Network (PGN) and BERT while also using our new VAE-PGN abstractive model. We evaluate our model on the benchmark Reddit corpus as part of the TL;DR challenge and show that our model outperforms the baseline in ROUGE score while generating diverse summaries.

Neural abstractive document summarization is commonly approached by models that exhibit a mostly extractive behavior. This behavior is facilitated by a copy-attention which allows models to copy words from a source document. While models in the mostly extractive news summarization domain benefit from this inductive bias, they commonly fail to paraphrase or compress information from the source document. Recent advances in transfer-learning from large pretrained language models give rise to alternative approaches that do not rely on copy-attention and instead learn to generate concise and abstractive summaries. In this paper, as part of the TL;DR challenge, we compare the abstractiveness of summaries from different summarization approaches and show that transfer-learning can be efficiently utilized without any changes to the model architecture. We demonstrate that the approach leads to a higher level of abstraction for a similar performance on the TL;DR challenge tasks, enabling true natural language compression.

In this paper, we report on the results of the TL;DR challenge, discussing an extensive manual evaluation of the expected properties of a good summary based on analyzing the comments provided by human annotators.

Quantified expressions have always taken up a central position in formal theories of meaning and language use. Yet quantified expressions have so far attracted far less attention from the Natural Language Generation community than, for example, referring expressions. In an attempt to start redressing the balance, we investigate a recently developed corpus in which quantified expressions play a crucial role; the corpus is the result of a carefully controlled elicitation experiment, in which human participants were asked to describe visually presented scenes. Informed by an analysis of this corpus, we propose algorithms that produce computer-generated descriptions of a wider class of visual scenes, and we evaluate the descriptions generated by these algorithms in terms of their correctness, completeness, and human-likeness. We discuss what this exercise can teach us about the nature of quantification and about the challenges posed by the generation of quantified expressions.

Generating grounded image descriptions requires associating linguistic units with their corresponding visual clues. A common method is to train a decoder language model with attention mechanism over convolutional visual features. Attention weights align the stratified visual features arranged by their location with tokens, most commonly words, in the target description. However, words such as spatial relations (e.g. next to and under) are not directly referring to geometric arrangements of pixels but to complex geometric and conceptual representations. The aim of this paper is to evaluate what representations facilitate generating image descriptions with spatial relations and lead to better grounded language generation. In particular, we investigate the contribution of three different representational modalities in generating relational referring expressions: (i) pre-trained convolutional visual features, (ii) different top-down geometric relational knowledge between objects, and (iii) world knowledge captured by contextual embeddings in language models.

In Natural Language Generation (NLG), End-to-End (E2E) systems trained through deep learning have recently gained a strong interest. Such deep models need a large amount of carefully annotated data to reach satisfactory performance. However, acquiring such datasets for every new NLG application is a tedious and time-consuming task. In this paper, we propose a semi-supervised deep learning scheme that can learn from non-annotated data and annotated data when available. It uses a NLG and a Natural Language Understanding (NLU) sequence-to-sequence models which are learned jointly to compensate for the lack of annotation. Experiments on two benchmark datasets show that, with limited amount of annotated data, the method can achieve very competitive results while not using any pre-processing or re-scoring tricks. These findings open the way to the exploitation of non-annotated datasets which is the current bottleneck for the E2E NLG system development to new applications.

We present the first dataset targeted at end-to-end NLG in Czech in the restaurant domain, along with several strong baseline models using the sequence-to-sequence approach. While non-English NLG is under-explored in general, Czech, as a morphologically rich language, makes the task even harder: Since Czech requires inflecting named entities, delexicalization or copy mechanisms do not work out-of-the-box and lexicalizing the generated outputs is non-trivial. In our experiments, we present two different approaches to this this problem: (1) using a neural language model to select the correct inflected form while lexicalizing, (2) a two-step generation setup: our sequence-to-sequence model generates an interleaved sequence of lemmas and morphological tags, which are then inflected by a morphological generator.

Recently, significant improvements have been achieved in various natural language processing tasks using neural sequence-to-sequence models. While aiming for the best generation quality is important, ultimately it is also necessary to develop models that can assess the quality of their output. In this work, we propose to use the similarity between training and test conditions as a measure for models’ confidence. We investigate methods solely using the similarity as well as methods combining it with the posterior probability. While traditionally only target tokens are annotated with confidence measures, we also investigate methods to annotate source tokens with confidence. By learning an internal alignment model, we can significantly improve confidence projection over using state-of-the-art external alignment tools. We evaluate the proposed methods on downstream confidence estimation for machine translation (MT). We show improvements on segment-level confidence estimation as well as on confidence estimation for source tokens. In addition, we show that the same methods can also be applied to other tasks using sequence-to-sequence models. On the automatic speech recognition (ASR) task, we are able to find 60% of the errors by looking at 20% of the data.

Deep neural networks (DNN) are quickly becoming the de facto standard modeling method for many natural language generation (NLG) tasks. In order for such models to truly be useful, they must be capable of correctly generating utterances for novel meaning representations (MRs) at test time. In practice, even sophisticated DNNs with various forms of semantic control frequently fail to generate utterances faithful to the input MR. In this paper, we propose an architecture agnostic self-training method to sample novel MR/text utterance pairs to augment the original training data. Remarkably, after training on the augmented data, even simple encoder-decoder models with greedy decoding are capable of generating semantically correct utterances that are as good as state-of-the-art outputs in both automatic and human evaluations of quality.

Variational Autoencoder (VAE) is a powerful method for learning representations of high-dimensional data. However, VAEs can suffer from an issue known as latent variable collapse (or KL term vanishing), where the posterior collapses to the prior and the model will ignore the latent codes in generative tasks. Such an issue is particularly prevalent when employing VAE-RNN architectures for text modelling (Bowman et al., 2016; Yang et al., 2017). In this paper, we present a new architecture called Full-Sampling-VAE-RNN, which can effectively avoid latent variable collapse. Compared to the general VAE-RNN architectures, we show that our model can achieve much more stable training process and can generate text with significantly better quality.