Lune

EMNLP Findings2024

Who Wrote When? Author Diarization in Social Media Discussions

Benedikt T. Boenninghoff, Henry Hosseini, Robert M. Nickel, Dorothea Kolossa

Abstract

We are proposing a novel framework for author diarization, i.e. attributing comments in online discussions to individual authors. We consider an innovative approach that merges pre-trained neural representations of writing style with author-conditional encoder-decoder diarization, enhanced by a Conditional Random Field with Viterbi decoding for alignment refinement. Additionally, we introduce two new large-scale German language datasets, one for authorship verification and the other for author diarization. We evaluate the performance of our diarization framework on these datasets, offering insights into the strengths and limitations of this approach.

Author Diarization (AD) tries to answer whether a single author or multiple authors have written a chronologically ordered series of texts. The task of author diarization can generally be described as the process of dividing an input text stream into text segments belonging to an individual author or to collaboratively working authors.

This work focuses on online discussions, such as those on a newspaper's forum or chat rooms. Two key assumptions guide our approach. Firstly, style changes are assumed to occur only between comments. Secondly, multiple authors are assumed to not collaborate on the same comment. Consequently, we redefine the task as follows: Given a sequence of comments, determine the number of authors and uniquely assign all discussion comments to their respective authors. Figure 1 illustrates the described scenario and the expected output for the author diarization task 1 .

Author diarization is intricately linked to the Authorship Verification (AV) and Style Change

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

Questions to start from

Lune can draw on

the full paper · related work

Get the answer

Free to start. No credit card required.

Lineage

Builds onCited byLeft to right by year; the more cited, the higher.

Builds on1

  1. Idiosyncratic but not Arbitrary: Learning Idiolects in Online Registers Reveals Distinctive yet Consistent Individual StylesJian Zhu, David JurgensEMNLP 2021 · 16 citations

Related papers

  1. Leveraging Multilingual Training for Authorship Representation: Enhancing Generalization across Languages and DomainsJunghwan Kim, Haotian Zhang, David JurgensEMNLP 2025 · 3 citations
  2. GerAV: Towards New Heights in German Authorship Verification using Fine-Tuned LLMs on a New BenchmarkLotta Kiefer, Christoph Leiter, Sotaro Takeshita, Elena Schmidt et al.ACL Findings 2026
  3. Dynamic Gaussian Embedding of AuthorsAntoine Gourru, Julien Velcin, Christophe Gravier, Julien JacquesWWW 2022 · 5 citations
  4. Adapting Language Models for Non-Parallel Author-Stylized RewritingBakhtiyar Syed, Gaurav Verma, Balaji Vasan Srinivasan, Anandhavelu Natarajan et al.AAAI 2020 · 53 citations
  5. Explainable Disentangled Representation Learning for Generalizable Authorship Attribution in the Era of Generative AIHieu Man, Van-Cuong Pham, Nghia Trung Ngo, Franck Dernoncourt et al.ACL 2026
Dusk over the sea between two cliffs drawn in fine vertical lines