Croatia

Authorship verification is used to link texts written by the same author without needing a model per author, making it useful for de-anonymizing users spreading text with malicious intent. Recent advances in Transformer-based language models hold great promise for author verification, though short context lengths and non-diverse training regimes present challenges for their practical application. In this work, we investigate the effect of these challenges in the application of a Cross-Encoder Transformer-based author verification system under multiple conditions. We perform experiments with four Transformer backbones using differently tuned variants of fanfiction data and found that our BigBird pipeline outperformed Longformer, RoBERTa, and ELECTRA and performed competitively against the official top ranked system from the PAN evaluation.
We also examined the effect of authors and fandoms not seen in training on model performance. Through this, we found fandom has the greatest influence on true trials, pairs of text written by the same author, and that a balanced training dataset in terms of class and fandom performed the most consistently.

EACL 2023

Improving Long-Text Authorship Verification via Model Selection and Data Tuning

author verification

large language models

transformers

workshop paper

### Welcome to the 17th Conference of the European Chapter of the Association for Computational Linguistics (EACL) 
Continuing its mission of expanding and involving the science community of all European countries, EACL had selected the Ukrainian community for the 16th EACL, which took place online due to the COVID pandemic. Unfortunately, the ongoing war made the organisation in Kyiv impossible. Considering the importance of physical interaction among researchers, especially after the restrictions imposed by the COVID pandemic, in addition to an online mode, the conference will be held in Dubrovnik, Croatia, from 2 to 6 of May, 2023. The original aim of strengthening the connection with the Ukrainian community will not change as our program will feature a dedicated session and a workshop to highlight work on Ukrainian language technologies. As the flagship European conference in the field of computational linguistics, EACL welcomes European and international researchers covering a broad spectrum of research areas that are concerned with computational approaches to natural language.

You need to register for the conference in order to access this site. Please visit https://2023.eacl.org/registration for more information.

As the flagship European conference in the field of computational linguistics, EACL welcomes European and international researchers covering a broad spectrum of research areas that are concerned with computational approaches to natural language.

We present a dataset and system for quote attribution in Dutch literature. The system is implemented as a neural module in an existing NLP pipeline for Dutch literature (dutchcoref; van Cranenburgh, 2019). Our contributions are as follows. First, we provide guidelines for Dutch quote attribution and annotate 3,056 quotes in fragments of 42 Dutch literary novels, both contemporary and classic. Second, we present three neural quote attribution classifiers, optimizing for precision, recall, and
F1. Third, we perform an evaluation and analysis of quote attribution performance, showing that in particular, quotes with an implicit speaker are challenging, and that such quotes are prevalent in contemporary fiction (57%, compared to 32% for classic novels). On the task of quote attribution, we achieve an improvement over the rule-based baseline of 8.0% F1 points on contemporary fiction and 1.9% F1 points on classic novels. Code, models, and annotations for the public domain novels are available under an open license at https://github.com/frenkvdberg/dutchqa

Direct Speech Quote Attribution for Dutch Literature

The subject of this article is the application of NLP and text-mining methods to the analysis of two large bibliographies: a Polish one, based on the catalogs of the National Library in Warsaw, and a German one, created by the Deutsche Nationalbibliothek. The data in both collections are stored in MARC 21 format, allowing the selection of relevant fields that are used for further processing (basically author, title, and date). The volume of the Polish corpus (after filtering out non-relevant or incomplete items) includes 1.4 mln of records, and that of the German corpus 7.5 mln records. The time span of both bibliographies extends from 1801 to 2021. The aim of the study is to compare the gender distribution of book authors in Polish and German databases over more than two centuries. The proportions of male and female authors since 1801 were calculated automatically, and NLP methods such as document vector embeddings based on deep BERT networks were used to extract topics from titles. The gender of the Polish authors was recognized based on the morphology of the first names, and that of the German authors based on a predefined list. The study found that the proportion of female authors has been steadily increasing both in Poland and in German countries (currently around 43%). However, the topics of women's and men's writings invariably remain different since 1801.

Large Bibliographies as a Source of Data for the Humanities – NLP in the Analysis of Gender of Book Authors in German Countries and in Poland (1801-2021)

Development funds are essential to finance climate change adaptation and are thus an important part of international climate policy. However, the absence of a common reporting practice makes it difficult to assess the amount and distribution of such funds. This problem has attracted attention in international affairs research and is increasingly being investigated using methods of the broader field of computational social science. Lately, the mentioned research field has questioned the credibility of
reported figures, indicating that adaptation financing is in fact lower than published figures suggest. Projects claiming a greater relevance to climate change adaptation than they target are referred to as "overreported". To estimate realistic rates of overreporting in large data sets over time, we propose an approach based on
state-of-the-art text classification. To date, assessments of credibility have relied on small, manually evaluated samples. We use such a sample data set to train a classifier with an accuracy of 89.81% ± 0.83% (tenfold crossvalidation) and extrapolate to larger data sets to identify overreporting. Additionally, we propose a method that incorporates evidence of smaller, higher-quality data to correct predicted rates using Bayes’ theorem. This enables a comparison of different annotation schemes to
estimate the degree of overreporting in climate change adaptation. Our results support findings that indicate extensive overreporting of 32.03% with a credible interval of [19.81%; 48.34%].

Using Text Classification with a Bayesian Correction for Estimating Overreporting in the Creditor Reporting System on Climate Adaptation Finance

The presence of bias is a clear and pressing concern for both engineers and users of language technology. What is less clear is how exactly bias can be measured, so as to rank models relative to the biases they display. Using an innovative experimental method involving data augmentation, we measure the effect of intersectional biases in Danish models used for Named Entity Recognition (NER). We quantify differences in representational biases, understood as a systematic difference in error or what is called error disparity. Our analysis includes both gender and ethnicity to illustrate the effect of multiple dimensions of bias, as well as experiments which look to move beyond a narrowly binary analysis of gender. We show that all contemporary Danish NER models perform systematically worse on non-binary and minority ethnic names, while not showing significant differences for typically Danish names. Our data augmentation technique can be applied on other languages to test for biases which might be relevant for researchers applying NER models to the study of textual cultural heritage data.

Detecting intersectionality in NER models: A data-driven approach

This paper presents a general-purpose NLP pipeline for Ancient or early forms of Greek (Classical, Koine, and Medieval) that achieves a slight state-of-art improvement by training on several Universal Dependencies treebanks jointly.
We measure the performance of the model against other comparable tools. We show that the selected Greek language models tend not to generalize well to out-of-training set samples. More work is necessary to ensure interoperability between the existing datasets. We identify the main issues and list suggestions for improvements.

OdyCy – A general-purpose NLP pipeline for Ancient Greek

This paper introduces the question answering paradigm as a way to explore digitized archive collections for Social Science studies.
Question generation can be used as a way to create explainable links between documents. Question generation for document linking is validated on a new corpus of digitized archive collection of a French Social Science journal.

Exploring Social Sciences Archives with Explainable Document Linkage through Question Generation

In social sciences, recent years have witnessed a growing interest in applying NLP approaches to automatically detect framing in political discourse. However, most NLP studies by now focus heavily on framing effect arising from topic coverage, whereas framing effect arising from subtle usage of linguistic devices remains
understudied. In a collaboration with political science researchers, we intend to investigate framing strategies in German newspaper articles on the "European Refugee Crisis". With the goal of a more in-depth framing analysis, we not only incorporate lexical cues for shallow topic-related framing, but also propose and operationalize a variety of framing-relevant semantic and pragmatic devices, which are theoretically derived from linguistics and political science research. We demonstrate the influential role of these linguistic devices with a large-scale quantitative analysis, bringing novel insights into the linguistic properties of framing.

Towards a More In-Depth Detection of Political Framing

This paper investigates the identification of populist rhetoric in text and presents a novel cross-lingual dataset for this task. Our work is based on the definition of populism as a "communication style of political actors that refers to the people" but also includes anti-elitism as another core feature of populism. Accordingly, we annotate references to people and elite in German and English parliamentary debates with a hierarchical scheme. The paper describes our dataset and annotation procedure and reports inter-annotator agreement for this task. Next, we compare and evaluate different transformer-based model architectures on a German dataset and report results for zero-shot learning on a smaller English data. We then show that semi-supervised tri-training can improve results in the cross-lingual setting. Our dataset can be used to investigate how political actors talk about elite and people and to study how populist rhetoric is used as a strategic device.

Our kind of people? Detecting populist references in political debates

Assessing the quality of an argument is a complex, highly subjective task, influenced by heterogeneous factors (e.g., prior beliefs of the an- notators, topic, domain, and application), and crucial for its impact in downstream tasks (e.g., argument retrieval or generation). Both the Argument Mining and the Social Science community have devoted plenty of attention to it, resulting in a wide variety of argument quality dimensions and a large number of annotated resources. This work aims at a better understanding of how the different aspects of argument quality relate to each other from a practical point of view. We employ adapter- fusion (Pfeiffer et al., 2021) as a multi-task learning framework which a) can improve the prediction of individual quality dimensions by injecting knowledge about related dimensions b) is efficient and modular and c) can serve as an analysis tool to investigate relations be- tween different dimensions. We conduct experiments on 6 datasets and 20 quality dimen- sions. We find that the majority of the dimensions can be learned as a weighted combination of other quality aspects, and that for 8 dimensions adapter fusion improves quality prediction. Last, we show the benefits of this approach by improving the performance in an extrinsic, out-of-domain task: prediction of mod- erator interventions in a deliberative forum.

Improving Long-Text Authorship Verification via Model Selection and Data Tuning

Downloads

Next from EACL 2023

Direct Speech Quote Attribution for Dutch Literature

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES