To convince readers of the novelty of their research paper, authors must perform a literature review and compose a coherent story that connects and relates prior works to the current work. This challenging nature of literature review writing makes automatic related work generation (RWG) academically and computationally interesting, and also makes it an excellent test bed for examining the capability of SOTA natural language processing (NLP) models. Since the initial proposal of the RWG task, its popularity has waxed and waned, following the capabilities of mainstream NLP approaches. In this work, we survey the zoo of RWG historical works, summarizing the key approaches and task definitions and discussing the ongoing challenges of RWG.

An automatic citation generation system aims to concisely and accurately describe the relationship between two scientific articles. To do so, such a system must ground its outputs to the content of the cited paper to avoid non-factual hallucinations. Due to the length of scientific documents, existing abstractive approaches have conditioned only on cited paper abstracts. We demonstrate empirically that the abstract is not always the most appropriate input for citation generation and that models trained in this way learn to hallucinate. We propose to condition instead on the cited text span (CTS) as an alternative to the abstract. Because manual CTS annotation is extremely time- and labor-intensive, we experiment with distant labeling of candidate CTS sentences, achieving sufficiently strong performance to substitute for expensive human annotations in model training, and we propose a human-in-the-loop, keyword-based CTS retrieval approach that makes generating citation texts grounded in the full text of cited papers both promising and practical.

Knowledge-based, open-domain dialogue generation aims to build chit-chat systems that talk to humans using mined support knowledge. Many types and sources of knowledge have previously been shown to be useful as support knowledge. Even in the era of large language models, response generation grounded in knowledge retrieved from additional up-to-date sources remains a practically important approach. While prior work using single-source knowledge has shown a clear positive correlation between the performances of knowledge selection and response generation, there are no existing multi-source datasets for evaluating support knowledge retrieval. Further, prior work has assumed that the knowledge sources available at test time are the same as during training. This unrealistic assumption unnecessarily handicaps models, as new knowledge sources can become available after a model is trained. In this paper, we present a high-quality benchmark named multi-source Wizard of Wikipedia (Ms.WoW) for evaluating multi-source dialogue knowledge selection and response generation. Unlike existing datasets, it contains clean support knowledge, grounded at the utterance level and partitioned into multiple knowledge sources. We further propose a new challenge, dialogue knowledge plug-and-play, which aims to test an already trained dialogue model on using new support knowledge from previously unseen sources in a zero-shot fashion.

Abstractive citation text generation is usually framed as an infilling task, where a sequence-to-sequence model is trained to generate a citation given a reference paper and the context window around the target; the generated citation should be a brief discussion of the reference paper as it relates to the citing context. However, examining a recent LED-based citation generation system, we find that many of the generated citations are generic summaries of the reference paper’s main contribution, ignoring the citation context’s focus on a different topic. To address this problem, we propose a simple modification to the citation text generation task: the generation target is not only the citation itself, but the entire context window, including the target citation. This approach can be easily applied to any abstractive citation generation system, and our experimental results show that training in this way is preferred by human readers and allows the generation model to make use of contextual clues about what topic to discuss and what stance to take.

Text segmentation is the task of dividing a sequence of text elements (eg. words, sentences, or paragraphs) into meaningful chunks. Although exciting advances are being made in modern segmentation-based tasks, such as automatically generating podcast chapters, current segmentation similarity metrics share a critical weakness: they are content-agnostic. In this paper, we present a word-embedding-based metric of cross-textual cohesion based on the formal linguistic definition of cohesion and incorporate it into a new segmentation similarity metric, SED. Our similarity metric, SED, is capable of providing fine-grained segmentation similarity scoring for the 3 basic segmentation errors: transposition, insertion, and deletion, as well as mixtures of them, avoiding the limitations of traditional metrics. We discuss the benefits of SED and evaluate its alignment with human judgement for each of the 3 basic error types. We show that our metric aligns with human evaluations significantly more than traditional metrics. We briefly discuss future work, such as the integration of anaphora resolution into our cohesion-based metric, and make our code publicly available.

Text segmentation is a natural language processing task with popular applications, such as topic segmentation, element discourse extraction, and sentence tokenization. Much work has been done to develop accurate segmentation similarity metrics, but even the most advanced metrics used today, B, and WindowDiff, exhibit incorrect behavior due to their evaluation of boundaries in isolation. In this paper, we present a new segment-alignment based approach to segmentation similarity scoring and a new similarity metric A. We show that A does not exhibit the erratic behavior of $ and WindowDiff, quantify the likelihood of B and WindowDiff misbehaving through simulation, and discuss the versatility of alignment-based approaches for segmentation similarity scoring. We make our implementation of A publicly available and encourage the community to explore more sophisticated approaches to text segmentation similarity scoring.

Academic research is an exploratory activity to discover new solutions to problems. By this nature, academic research works perform literature reviews to distinguish their novelties from prior work. In natural language processing, this literature review is usually conducted under the “Related Work” section. The task of related work generation aims to automatically generate the related work section given the rest of the research paper and a list of papers to cite. Prior work on this task has focused on the sentence as the basic unit of generation, neglecting the fact that related work sections consist of variable length text fragments derived from different information sources. As a first step toward a linguistically-motivated related work generation framework, we present a Citation Oriented Related Work Annotation (CORWA) dataset that labels different types of citation text fragments from different information sources. We train a strong baseline model that automatically tags the CORWA labels on massive unlabeled related work section texts. We further suggest a novel framework for human-in-the-loop, iterative, abstractive related work generation.

Social media has changed the way we engage in social activities. On Twitter, users can participate in social movements using hashtags such as #MeToo; this is known as hashtag activism. However, while these hashtags can help reshape social norms, they can also be used maliciously by spammers or troll communities for other purposes, such as signal boosting unrelated content, making a dent in a movement, or sharing hate speech. We present a Tweet-level hashtag hijacking detection framework focusing on hashtag activism. Our weakly-supervised framework uses bootstrapping to update itself as new Tweets are posted. Our experiments show that the system adapts to new topics in a social movement, as well as new hijacking strategies, maintaining strong performance over time.

Human language encompasses more than just text; it also conveys emotions through tone and gestures. We present a case study of three simple and efficient Transformer-based architectures for predicting sentiment and emotion in multimodal data. The Late Fusion model merges unimodal features to create a multimodal feature sequence, the Round Robin model iteratively combines bimodal features using cross-modal attention, and the Hybrid Fusion model combines trimodal and unimodal features together to form a final feature sequence for predicting sentiment. Our experiments show that our small models are effective and outperform the publicly released versions of much larger, state-of-the-art multimodal sentiment analysis systems.

We present a monolingual alignment system for long, sentence- or clause-level alignments, and demonstrate that systems designed for word- or short phrase-based alignment are ill-suited for these longer alignments. Our system is capable of aligning semantically similar spans of arbitrary length. We achieve significantly higher recall on aligning phrases of four or more words and outperform state-of-the- art aligners on the long alignments in the MSR RTE corpus.

We present a robust neural abstractive summarization system for cross-lingual summarization. We construct summarization corpora for documents automatically translated from three low-resource languages, Somali, Swahili, and Tagalog, using machine translation and the New York Times summarization corpus. We train three language-specific abstractive summarizers and evaluate on documents originally written in the source languages, as well as on a fourth, unseen language: Arabic. Our systems achieve significantly higher fluency than a standard copy-attention summarizer on automatically translated input documents, as well as comparable content selection.

We present an iterative annotation process for producing aligned, parallel corpora of abstractive and extractive summaries for narrative. Our approach uses a combination of trained annotators and crowd-sourcing, allowing us to elicit human-generated summaries and alignments quickly and at low cost. We use crowd-sourcing to annotate aligned phrases with the text-to-text generation techniques needed to transform each phrase into the other. We apply this process to a corpus of 476 personal narratives, which we make available on the Web.

We present novel computational experiments using William Labov’s theory of narrative analysis. We describe his six elements of narrative structure and construct a new corpus based on his most recent work on narrative. Using this corpus, we explore the correspondence between Labov’s elements of narrative structure and the implicit discourse relations of the Penn Discourse Treebank, and we construct a mapping between the elements of narrative structure and the discourse relation classes of the PDTB. We present first experiments on detecting Complicating Actions, the most common of the elements of narrative structure, achieving an f-score of 71.55. We compare the contributions of features derived from narrative analysis, such as the length of clauses and the tenses of main verbs, with those of features drawn from work on detecting implicit discourse relations. Finally, we suggest directions for future research on narrative structure, such as applications in assessing text quality and in narrative generation.