In this paper, we show how three often used and seemingly different discourse annotation frameworks – Penn Discourse Treebank (PDTB), Rhetorical Structure Theory (RST), and Segmented Discourse Representation Theory – can be related by using a set of unifying dimensions. These dimensions are taken from the Cognitive approach to Coherence Relations and combined with more fine-grained additional features from the frameworks themselves to yield a posited set of dimensions that can successfully map three frameworks. The resulting interface will allow researchers to find identical or at least closely related relations within sets of annotated corpora, even if they are annotated within different frameworks. Furthermore, we tested our unified dimension (UniDim) approach by comparing PDTB and RST annotations of identical newspaper texts and converting their original end label annotations of relations into the accompanying values per dimension. Subsequently, rates of overlap in the attributed values per dimension were analyzed. Results indicate that the proposed dimensions indeed create an interface that makes existing annotation systems “talk to each other.”

Fake news has become an important topic of research in a variety of disciplines including linguistics and computer science. In this paper, we explain how the problem is approached from the perspective of natural language processing, with the goal of building a system to automatically detect misinformation in news. The main challenge in this line of research is collecting quality data, i.e., instances of fake and real news articles on a balanced distribution of topics. We review available datasets and introduce the MisInfoText repository as a contribution of our lab to the community. We make available the full text of the news articles, together with veracity labels previously assigned based on manual assessment of the articles' truth content. We also perform a topic modelling experiment to elaborate on the gaps and sources of imbalance in currently available datasets to guide future efforts. We appeal to the community to collect more data and to make it available for research purposes.

Discourse-annotated corpora are an important resource for the community, but they are often annotated according to different frameworks. This makes joint usage of the annotations difficult, preventing researchers from searching the corpora in a unified way, or using all annotated data jointly to train computational systems. Several theoretical proposals have recently been made for mapping the relational labels of different frameworks to each other, but these proposals have so far not been validated against existing annotations. The two largest discourse relation annotated resources, the Penn Discourse Treebank and the Rhetorical Structure Theory Discourse Treebank, have however been annotated on the same texts, allowing for a direct comparison of the annotation layers. We propose a method for automatically aligning the discourse segments, and then evaluate existing mapping proposals by comparing the empirically observed against the proposed mappings. Our analysis highlights the influence of segmentation on subsequent discourse relation labelling, and shows that while agreement between frameworks is reasonable for explicit relations, agreement on implicit relations is low. We identify several sources of systematic discrepancies between the two annotation schemes and discuss consequences for future annotation and for usage of the existing resources.

Word embeddings obtained from neural network models such as Word2Vec Skipgram have become popular representations of word meaning and have been evaluated on a variety of word similarity and relatedness norming data. Skipgram generates a set of word and context embeddings, the latter typically discarded after training. We demonstrate the usefulness of context embeddings in predicting asymmetric association between words from a recently published dataset of production norms (Jouravlev and McRae, 2016). Our findings suggest that humans respond with words closer to the cue within the context embedding space (rather than the word embedding space), when asked to generate thematically related words.

We examine gender bias in media by tallying the number of men and women quoted in news text, using the Gender Gap Tracker, a software system we developed specifically for this purpose. The Gender Gap Tracker downloads and analyzes the online daily publication of seven English-language Canadian news outlets and enhances the data with multiple layers of linguistic information. We describe the Natural Language Processing technology behind this system, the curation of off-the-shelf tools and resources that we used to build it, and the parts that we developed. We evaluate the system in each language processing task and report errors using real-world examples. Finally, by applying the Tracker to the data, we provide valuable insights about the proportion of people mentioned and quoted, by gender, news organization, and author gender. Data collected between October 1, 2018 and September 30, 2020 shows that, in general, men are quoted about three times as frequently as women. While this proportion varies across news outlets and time intervals, the general pattern is consistent. We believe that, in a world with about 50% women, this should not be the case. Although journalists naturally need to quote newsmakers who are men, they also have a certain amount of control over who they approach as sources. The Gender Gap Tracker relies on the same principles as fitness or goal-setting trackers: By quantifying and measuring regular progress, we hope to motivate news organizations to provide a more diverse set of voices in their reporting.