@Book{DTF:2026,
  editor    = {Barth, Florian; Du, Keli; Calvo Tello, José; Genêt, Philippe; Lendvai, Piroska; Schöch, Christof; Trippel, Thorsten},
  title     = {Proceedings of Leveraging Derived Text Formats to Unlock Copyrighted Collections for Open Science (DTF) @ LREC 2026},
  month          = {June},
  year           = {2026},
  address        = {Palma, Mallorca (Spain)},
  publisher      = {ELRA Language Resources Association (ELRA)},
  url       = {https://aclanthology.org/2026.dtf-1}
}

@InProceedings{schch:2026:DTF,
  author    = {Schöch, Christof},
  title     = {Derived Text Formats as Strategic Transformations of In-Copyright Materials to Support Open Science: A Survey},
  booktitle      = {Proceedings of Leveraging Derived Text Formats to Unlock Copyrighted Collections for Open Science (DTF) @ LREC 2026},
  month          = {June},
  year           = {2026},
  address        = {Palma, Mallorca (Spain)},
  publisher      = {ELRA Language Resources Association (ELRA)},
  pages     = {1--15},
  abstract  = {Derived Text Formats (DTFs) are the result of a strategic transformation of textual materials that are protected by copyright in their original form, such that the resulting data is useful for computational analyses and can be openly shared following best practices of Open Science without infringing copyright law. This paper aims to provide insights into several key aspects of this concept that is closely related to concepts such as corpus masking, non-consumptive research and extracted features. The paper establishes the motivation for using DTFs, discusses several foundational aspects of the concept and practice, describes ongoing research on issues including copyright, reconstructibility, evaluation and standardization of DTFs, and concludes with a roadmap for future work on DTFs. In this way, this paper provides a broad but concise overview of work on DTFs as a contribution to Open Science practices, with a focus on work in the Digital Humanities.},
  url       = {https://aclanthology.org/2026.dtf-1.1}
}

@InProceedings{du-schch:2026:DTF1,
  author    = {Du, Keli  and  Schöch, Christof},
  title     = {A Multi-dimensional Constrained Framework for Derived Text Formats},
  booktitle      = {Proceedings of Leveraging Derived Text Formats to Unlock Copyrighted Collections for Open Science (DTF) @ LREC 2026},
  month          = {June},
  year           = {2026},
  address        = {Palma, Mallorca (Spain)},
  publisher      = {ELRA Language Resources Association (ELRA)},
  pages     = {16--19},
  abstract  = {Derived Text Formats (DTFs) have been proposed as a solution to enable text and data mining while avoiding copyright infringement. Building on a review of recent empirical studies of DTFs on topic modeling, authorship classification, and sentiment analysis, this paper argues that DTFs should not be treated as static formats, but as variable and task-dependent representations shaped by multiple interacting factors. In response, we propose a multi-dimensional framework that conceptualizes DTFs as configurations within a structured space defined by both internal representation parameters and external constraints. The framework includes four internal representation dimensions—feature level, degree of reduction, transformation strategy, and aggregation level—as well as two external constraining forces: legal requirements and task-specific information needs. By emphasizing the interdependence of these dimensions, the proposed framework provides a systematic way to describe, compare, and design DTFs across different analytical contexts. Therefore, this paper contributes to a more theoretically grounded understanding of DTFs and offers guidance for their responsible and effective use in text and data mining in Digital Humanities.},
  url       = {https://aclanthology.org/2026.dtf-1.2}
}

@InProceedings{iacino-kamocki-du:2026:DTF,
  author    = {Iacino, Gianna  and  Kamocki, Pawel  and  Du, Keli},
  title     = {Legal implications of Derived Text Formats - a copyright perspective},
  booktitle      = {Proceedings of Leveraging Derived Text Formats to Unlock Copyrighted Collections for Open Science (DTF) @ LREC 2026},
  month          = {June},
  year           = {2026},
  address        = {Palma, Mallorca (Spain)},
  publisher      = {ELRA Language Resources Association (ELRA)},
  pages     = {20--24},
  abstract  = {Text and Data Mining (TDM) methods are often used in order to analyse large amounts of text for scientific research. If the analysed text is protected by copyright, the use of such TDM methods has copyright implications. The existing copyright exceptions facilitate TDM within a narrow framework which limits the storage, publication and re-use of datasets. This paper examines the legal framework of converting the source text into a derived text format (DTF) which is no longer protected by copyright in order to allow the use of TDM without legal restrictions. First, the creation itself of a DTF is being examined: it entails copyright relevant acts which are covered by the TDM exception. In a second step the copyright status of the created DTF has to be evaluated based on three criteria: the DTF may not contain elements which are an expression of the intellectual creation of the author of the source material, the source material may not be easily reconstructable based on the DTF and the source material may not be recognizable.},
  url       = {https://aclanthology.org/2026.dtf-1.3}
}

@InProceedings{rehm-trippel-witt:2026:DTF,
  author    = {Rehm, Georg  and  Trippel, Thorsten  and  Witt, Andreas},
  title     = {Revisiting Masking After Fifteen Years: Early Approaches to Non-Reconstructable Linguistic Data in the current context},
  booktitle      = {Proceedings of Leveraging Derived Text Formats to Unlock Copyrighted Collections for Open Science (DTF) @ LREC 2026},
  month          = {June},
  year           = {2026},
  address        = {Palma, Mallorca (Spain)},
  publisher      = {ELRA Language Resources Association (ELRA)},
  pages     = {25--33},
  abstract  = {This paper revisits the masking approaches introduced in 2007 for enabling the distribution of linguistically annotated corpora without exposing copyrighted or sensitive source texts and situates them within the contemporary framework of Derived Text Formats (DTF). While the original work demonstrated how syntactic and morphological information could be preserved through parameterised masking, today's landscape, which is shaped by large language models, FAIR requirements, and emerging standardisation efforts, demands more formalised, robust and reproducible methods. We outline how DTF extend early masking concepts by introducing explicit abstraction levels, reversibility classes, and machine- actionable provenance, supported by standards such as TEI, ISO linguistic annotation models, CMDI metadata, and the draft DIN DTF specification. Building on these foundations, we present a modern workflow for DTF generation, including enrichment pipelines, structural abstractions, statistical and embedding-based representations, and non-reversible transformation layers, illustrated through the MONA-pipe framework. We conclude that DTF constitute a sustainable and infrastructure - ready solution for open, reproducible and legally secure text-based research in the decades to come.},
  url       = {https://aclanthology.org/2026.dtf-1.4}
}

@InProceedings{ecker-schneider:2026:DTF,
  author    = {Ecker, Jennifer  and  Schneider, Roman},
  title     = {Multi-Label Text Classification of Derived Text Formats with DistilBERT},
  booktitle      = {Proceedings of Leveraging Derived Text Formats to Unlock Copyrighted Collections for Open Science (DTF) @ LREC 2026},
  month          = {June},
  year           = {2026},
  address        = {Palma, Mallorca (Spain)},
  publisher      = {ELRA Language Resources Association (ELRA)},
  pages     = {34--43},
  abstract  = {Derived Text Formats enable the distribution of copyrighted texts by systematically perturbing linguistic information to reduce reconstructability. However, the extent to which such information loss affects downstream text classification remains unclear. We investigate how controlled perturbations affect learning dynamics in transformer-based classification using two datasets and two strategies: POS-consistent replacement of 30\%, 40\%, and 50\% of tokens, and random word-order shuffling. On Wikipedia data, POS replacement increases loss by 4-9\% and reduces micro-F1 by 3-8\%, depending on the replacement rate, while shuffling raises loss by 5\% and lowers micro-F1 by 4\%. Performance degrades monotonically with higher replacement rates, and shuffling yields results between the 30\% and 40\% conditions, indicating that DistilBERT relies more on lexical semantics than on word order. Experiments on specialist-domain data show the same pattern, demonstrating robustness across domains. To test cross-representation generalization, we train classifiers on both clean and perturbed texts and evaluate them on the respective alternate representation. Models trained on DTF data generalize better to clean text than vice versa, suggesting that perturbation-based training promotes more robust representations. Our findings position DTF as a promising strategy for reproducible, legally compliant, and robust NLP research.},
  url       = {https://aclanthology.org/2026.dtf-1.5}
}

@InProceedings{indel-EtAl:2026:DTF,
  author    = {Šindelář, Pavel  and  Prášil, Filip  and  Slivka, Dávid  and  Bouma, Christopher  and  Bojar, Ondrej},
  title     = {Training data generation for context-dependent rubric-based short answer grading},
  booktitle      = {Proceedings of Leveraging Derived Text Formats to Unlock Copyrighted Collections for Open Science (DTF) @ LREC 2026},
  month          = {June},
  year           = {2026},
  address        = {Palma, Mallorca (Spain)},
  publisher      = {ELRA Language Resources Association (ELRA)},
  pages     = {44--50},
  abstract  = {Every four years, the PISA test is administered by the OECD to test the knowledge of teenage students worldwide and allow for comparisons of educational systems. However, having to avoid language differences and annotator bias makes the grading of student answers challenging. For these reasons, it would be interesting to consider methods of automatic student answer grading. To train some of these methods, which require machine learning, or to compute parameters or select hyperparameters for those that do not, a large amount of domain-specific data is needed. In this work, we explore a small number of methods for creating a large-scale training dataset using only a relatively small confidential dataset as a reference, leveraging a set of very simple derived text formats to preserve confidentiality. Using the proposed methods, we successfully created three surrogate datasets that are, at the very least, superficially more similar to the reference dataset than a straightforward result of prompt-based generation. Early experiments suggest one of these approaches might also lead to improved training of automatic answer grading models.},
  url       = {https://aclanthology.org/2026.dtf-1.6}
}

@InProceedings{laguidi-EtAl:2026:DTF,
  author    = {Laâguidi, Jammila  and  Ruban, Vitaliia  and  Laarmann-Quante, Ronja  and  Drackert, Anastasia},
  title     = {DUO\_DE A1: An Annotated Corpus of Online Learning Material for Beginning Learners of German as a Foreign Language},
  booktitle      = {Proceedings of Leveraging Derived Text Formats to Unlock Copyrighted Collections for Open Science (DTF) @ LREC 2026},
  month          = {June},
  year           = {2026},
  address        = {Palma, Mallorca (Spain)},
  publisher      = {ELRA Language Resources Association (ELRA)},
  pages     = {51--62},
  abstract  = {This paper describes the creation of DUO\_DE A1, a corpus based on A1-level learning material from the Deutsch-Uni Online (DUO) language courses for German as a foreign language. We split the material into small segments and manually annotated each with fine-grained information such as the type of segment (e.g. task description, description of grammar), the medium (e.g. text, table, audio), the text units it contains (e.g. words, phrases, sentences) and other special features (e.g. marking cloze texts). Furthermore, we automatically tokenized, POS tagged and lemmatized the corpus and compared the performance of three models on these steps for different kinds of segments. We publish the created corpus in a manner that respects copyright, releasing all structural features, metadata and POS tags.},
  url       = {https://aclanthology.org/2026.dtf-1.7}
}

@InProceedings{du-schch:2026:DTF2,
  author    = {Du, Keli  and  Schöch, Christof},
  title     = {Why Reconstructing Scrambled Texts Fails},
  booktitle      = {Proceedings of Leveraging Derived Text Formats to Unlock Copyrighted Collections for Open Science (DTF) @ LREC 2026},
  month          = {June},
  year           = {2026},
  address        = {Palma, Mallorca (Spain)},
  publisher      = {ELRA Language Resources Association (ELRA)},
  pages     = {63--66},
  abstract  = {This paper explores the limitations of reconstructing scrambled text within the context of Derived Text Formats (DTFs). While previous research has treated reconstruction as a technical challenge, this study shifts the focus to investigating the causes of reconstruction failure. Through a detailed analysis of outputs generated by language models on non-literary (IMDb reviews) and literary (Gutenberg texts) datasets, several systematic patterns were identified. First, reconstructed texts are generally shorter than the originals, indicating that the generated results are often incomplete. Second, models simplify expressions by omitting specific modifiers, thereby producing more general outputs. Third, high similarity at the string level does not guarantee semantic equivalence, revealing fidelity-related issues in text reconstruction. In literary texts, chunk-based segmentation poses additional challenges; this approach disrupts syntactic and contextual coherence, leading to sentences that are structurally correct but semantically distorted. These findings suggest that reconstruction difficulty is not merely a matter of model performance but also reflects the importance of higher-level textual organization. This study highlights the fundamental limitations of current language models and reframes reconstruction failure as an analytical perspective for understanding how meaning is constructed in text.},
  url       = {https://aclanthology.org/2026.dtf-1.8}
}

@InProceedings{trippel-EtAl:2026:DTF,
  author    = {Trippel, Thorsten  and  Barth, Florian  and  Tello, Jose Calvo  and  Du, Keli  and  Genêt, Philippe  and  Kurzawe, Daniel  and  Leinen, Peter  and  Lendvai, Piroska  and  Schöch, Christof  and  Witt, Andreas  and  Zimmermann, Arden},
  title     = {DIN 19461: A National Standard for Derived Text Formats},
  booktitle      = {Proceedings of Leveraging Derived Text Formats to Unlock Copyrighted Collections for Open Science (DTF) @ LREC 2026},
  month          = {June},
  year           = {2026},
  address        = {Palma, Mallorca (Spain)},
  publisher      = {ELRA Language Resources Association (ELRA)},
  pages     = {67--75},
  abstract  = {We present DIN~19461:2026-05 (E), a German draft national standard that defines categories, terminology, and process requirements for Derived Text Formats (DTFs) created from text documents in natural language. The standard specifies enrichment and information reduction operations, requirements for combining multiple DTFs, and documentation obligations for publication, archiving, and reuse. Its aim is to enable legally compliant sharing and analysis of texts--especially where copyright or data protection prevents distributing originals--while maintaining scientific utility and reproducibility through explicit process and parameter recording. We outline the scope, the key concepts, the four core reduction operations (retain, delete, replace, randomise), together with examples across token-, structure-, and vector-based DTFs, and implications for infrastructures (e.g., ISO 24622-based metadata). Finally, we discuss limitations, open questions (e.g., reconstruction risks with modern ML models), and next steps for adoption and maintenance.},
  url       = {https://aclanthology.org/2026.dtf-1.9}
}