13th Web-as-Corpus Workshop

The 13th edition of the Web-as-Corpus (WaC-13) workshop will be co-located with EMNLP 2026 in Budapest, Hungary. The workshop is endorsed by the SIGWAC ACL Special Interest Group in Web-as-Corpus, the Common Crawl Foundation, and the OpenEuroLLM project.

Call for Papers

The World Wide Web has evolved from a resource for building linguistic corpora into the central data infrastructure powering modern natural language processing and Large Language Models (LLMs). As web-scale data increasingly shapes AI systems' knowledge and capabilities, understanding its quality, representativeness, and ethical implications has become critical.

At the same time, the "more is better" paradigm is being challenged by issues such as machine-generated content, data toxicity, limited metadata, and the under-representation of many languages and domains. These challenges call for a shift toward Data-Centric AI, focusing on the curation, analysis, and responsible use of web-derived data.

WaC-13 provides a multidisciplinary forum for research addressing the full lifecycle of web data. We invite submissions on methods, resources, and applications related to web corpora, with special emphasis on multilingual data and less-resourced languages.

Programme

Papers are published in the Proceedings of the 13th Web-as-Corpus Workshop on the ACL Anthology (full volume PDF).

Thursday 29 October 2026 · Room G6 · All times are Budapest local time (CET, UTC+1)

1. 09:00–09:15
   Opening
2. 09:15–10:30
   Keynote

   The Web as a Corpus and Common Crawl: A look back, forward and into present challenges
   Sebastian Nagel, Common Crawl Foundation
3. 10:30–11:00
   Coffee break
4. 11:00–12:30
   Oral session 1

   1. 11:00–11:20Long oral Wiktionary as a Crowdsourced Lexicon for English DialectsremoteSidney Wong
   2. 11:20–11:40Long oral Host-Aware Web Crawling for LLM Pretraining DataMalte Ostendorff
   3. 11:40–11:55Short oral CHALIS: A Challenge Dataset for Language Identification in Difficult ScenariosMichal Tichý and Jindřich Libovický
   4. 11:55–12:10Short oral You Can Tell Who’s Asking: What the Web’s Questions Are Made Of, and Where They Come FromremoteCalvin Zhou, Vincent McCloskey and Krishna Srinivasan
   5. 12:10–12:30Long oral Segment-Level Register Analysis of the Multilingual WebErik Henriksson, Alireza Razzaghi, Tomasz Galica, Jiaqi Feng Guo, Anastasia Zhi-Ling Dong, Urho Kunnas, Erofili Psaltaki, Tuomas Lundberg, Antti Kanner and Veronika Laippala
5. 12:30–13:30
   Lunch break
6. 13:30–14:00
   Virtual poster presentations

   1. 13:30–13:35Lightning talk Who Published This Headline? Editorial Fingerprinting in French News Under Topic ControlremoteAmr Sobhy
   2. 13:35–13:40Recorded talk How Many Duplicates Are Left? Capture-Recapture and Sampling Inversion for Residual Duplication in FineWebGary Zhang
   3. 13:40–13:45Lightning talk Auditing Source Exposure in Baidu and Google AI SearchremoteYibo Li, Enci Guan, Yuedan Cai, Geng Liu and Francesco Pierri
   4. 13:45–13:50Recorded talk Same Pages, Cleaner Labels: Language Identification for Eight Low-Resource Languages in mC4 and FineWeb-2Gary Zhang
   5. 13:50–13:55Lightning talk Extraction Changes the Route: Main-Content Extraction and Language Identification in Web CorporaremoteToprak Seda Karaosmanoğlu
   6. 13:55–14:00Lightning talk WiNCC: A Reproducible Pipeline for Constructing a Wikipedia-enriched Nepali Cross-lingual CorpusremoteAhrnish P Dahal, Pushkar Baranwal, Amit Pundir, Geetika Jain Saxena and Sanjeev Singh
7. 14:00–15:30
   In-person poster session

   16 posters

   • A Web-Derived Typology of Multimodal Infant Crying: Taxonomy, a Text-Only Web-Corpus Comparison, and Ethical IssuesJacob Rigal and Jonathan Ginzburg
   • A Taxonomy of Programming Languages for Code GenerationNishat Raihan, Christian Newman and Marcos Zampieri
   • Can an Agent Debug a Web Corpus? Agentic Auditing of Slovak FineWeb2non-archivalMarek Šuppa, Ivan Vykopal and Jaroslav Kopcan
   • CeyNews: A Sri Lankan Trilingual News CorpusAjintha Sivakulasingam, Kengatharaiyer Sarveswaran and Tharindu Ranasinghe
   • Counting Documents Is Not Counting Text: Unit Bias in Web-PDF Corpus StatisticsLuca Foppiano
   • Is Educational Content Enough? The Linguistic and Thematic Effects of Educational FilteringUrho Kunnas, Veronika Laippala, Sampo Pyysalo and Tomasz Galica
   • Keeping the Ums: Conversation-Analysis-Informed Verbatim ASR for Web SpeechSteven Coats
   • Mapping Sámi Languages in a Norwegian National Web Crawl: Results from a Representativeness StudyMagnus Breder Birkenes
   • PAWN++ in the Wild: Evaluating AI-Generated Text Detection on the WebOleg Chernikov, Evgenii Ptitsyn and Albina Burlova
   • Probing Pretraining “Quality”: Sociolinguistic Mobility as a Lens on Web Corpus FiltersJessica Wang
   • Task-Agnostic Web Document Annotation with LLM-Generated DescriptorsOtto Tarkka, Erik Henriksson, Jenna Kanerva and Filip Ginter
   • The Crawl Gap: Where Endangered-Language Web Content Is Lost Before It Reaches a CorpusSharvin Goyal
   • The CreateDebate Arguments Dataset: A Large-Scale Naturalistic Corpus for Computational ArgumentationAntonis Charalampous and Constantinos Djouvas
   • Towards More Reliable Web Metadata: Do Register Classifiers Learn Register or Genre Features?Antti Kanner, Tuomas Lundberg, Erik Henriksson and Veronika Laippala
   • Using Search Engines to Trace Metalinguistic Discussions of the Noun “Female” in Digital DiscourseMarie Flesch
   • What Persists across Web Crawls? Genre and News Topic Variation in South Slavic Web CorporaNikola Ljubešić, Vuk Dinić and Peter Rupnik
8. 15:30–16:00
   Coffee break
9. 16:00–17:00
   Oral session 2

   1. 16:00–16:20Long oral RefinedGAW: A Refined German Academic Web Datasetnon-archivalMaximilian Idahl, Oleh Astappiev, Benedikt Droste, Robert Scholz, Björn Plüster, Jan Philipp Harries, Robert Jäschke and Wolfgang Nejdl
   2. 16:20–16:35Short oral Invisible on the Web: An Audit of Low-Resource Muskogean Languages in Web-Crawled CorporaSharvin Goyal
   3. 16:35–16:50Short oral HPLT3-PerRef: A Refined Large-Scale Persian Web Corpus via Supervised Document Quality FilteringAlireza Razzaghi, Tomasz Galica, Erik Henriksson and Veronika Laippala
   4. 16:50–16:55Lightning talk Lost in Translation: Structured-Span Corruption in Machine-Translated Multilingual Web CorporaremoteRuiyang Zhang
   5. 16:55–17:00Recorded talk Whose Robots.txt? Consent-Infrastructure Inequality Across the Languages of Web CorporaGary Zhang and Austin Cheng
10. 17:00–17:15
    Best paper award & wrap-up

Topics of Interest

Submissions are invited on (but not limited to):

• Creation and evaluation of high-quality datasets for foundation models (data collection, filtering, enrichment, language identification)
• Use of web data in empirical linguistic research
• Analysis of web-scale corpora for quality, representativeness, and societal insights
• Ethical and legal aspects of collecting, sharing, and using web data

Important Dates

Submission

We invite short (4 pages) and long (8 pages) papers. We allow for archival (to appear in the ACL Anthology) and non-archival submissions. Submissions should follow the ARR submission requirements. Following the ACL and ARR policies, there is no anonymity period requirement.

Paper submissions must use the official ACL style template. Please follow the paper formatting guidelines general to “*ACL” conferences. Authors may not modify these style files or use template designed for other conferences. The LaTeX style files are also available as an Overleaf template.

Submissions will be possible through ARR commitment and through direct submission via openreview.net

Workshop Organizers