@inproceedings{Lenz2026NumbersDontLie,
  title = {Numbers {{Don}}'t {{Lie}}: {{Hybrid Extraction}} and~{{Validation}} of~{{Quantitative Statements}} in~{{Arguments}} with~{{Semi-structured Information}}},
  shorttitle = {Numbers {{Don}}'t {{Lie}}},
  booktitle = {{{KI}} 2025: {{Advances}} in {{Artificial Intelligence}}},
  author = {Lenz, Mirko and Dumani, Lorik and Schenkel, Ralf and Bergmann, Ralph},
  editor = {Braun, Tanya and Paa{\ss}en, Benjamin and Stolzenburg, Frieder},
  year = {2026},
  series = {Lecture {{Notes}} in {{Computer Science}}},
  volume = {15956},
  pages = {77--90},
  publisher = {Springer Nature Switzerland},
  address = {Cham},
  doi = {10.1007/978-3-032-02813-6_6},
  abstract = {Evidence in arguments may be stated in various forms, including quantitative statements (i.e., numerical relations between entities). This measurable information can be validated against reliable sources like Wikipedia to combat the spread of misinformation. In this paper, we propose a four-step pipeline that combines rule-based techniques with prompting strategies for generative language models in a hybrid fashion. We use regular expressions to identify candidates in claim-premise structures, extract statements using GPT-4o, augment the data with tables from Wikipedia, and validate statements through retrieval-augmented generation (RAG). The pipeline is evaluated on two existing argumentation corpora and the generated dataset is manually annotated to assess the quality of our predictions, showing promising results for extraction and mixed results for validation. Our code and data are available to foster further research in this area.},
  isbn = {978-3-032-02813-6},
  langid = {english},
  url = {https://www.wi2.uni-trier.de/shared/publications/Lenz2026NumbersDontLie.pdf}
}

{"_id":"f7AQDfjSCPHLf4K6h","bibbaseid":"lenz-dumani-schenkel-bergmann-numbersdontliehybridextractionandvalidationofquantitativestatementsinargumentswithsemistructuredinformation-2026","author_short":["Lenz, M.","Dumani, L.","Schenkel, R.","Bergmann, R."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","title":"Numbers Don't Lie: Hybrid Extraction and Validation of Quantitative Statements in Arguments with Semi-structured Information","shorttitle":"Numbers Don't Lie","booktitle":"KI 2025: Advances in Artificial Intelligence","author":[{"propositions":[],"lastnames":["Lenz"],"firstnames":["Mirko"],"suffixes":[]},{"propositions":[],"lastnames":["Dumani"],"firstnames":["Lorik"],"suffixes":[]},{"propositions":[],"lastnames":["Schenkel"],"firstnames":["Ralf"],"suffixes":[]},{"propositions":[],"lastnames":["Bergmann"],"firstnames":["Ralph"],"suffixes":[]}],"editor":[{"propositions":[],"lastnames":["Braun"],"firstnames":["Tanya"],"suffixes":[]},{"propositions":[],"lastnames":["Paaßen"],"firstnames":["Benjamin"],"suffixes":[]},{"propositions":[],"lastnames":["Stolzenburg"],"firstnames":["Frieder"],"suffixes":[]}],"year":"2026","series":"Lecture Notes in Computer Science","volume":"15956","pages":"77–90","publisher":"Springer Nature Switzerland","address":"Cham","doi":"10.1007/978-3-032-02813-6_6","abstract":"Evidence in arguments may be stated in various forms, including quantitative statements (i.e., numerical relations between entities). This measurable information can be validated against reliable sources like Wikipedia to combat the spread of misinformation. In this paper, we propose a four-step pipeline that combines rule-based techniques with prompting strategies for generative language models in a hybrid fashion. We use regular expressions to identify candidates in claim-premise structures, extract statements using GPT-4o, augment the data with tables from Wikipedia, and validate statements through retrieval-augmented generation (RAG). The pipeline is evaluated on two existing argumentation corpora and the generated dataset is manually annotated to assess the quality of our predictions, showing promising results for extraction and mixed results for validation. Our code and data are available to foster further research in this area.","isbn":"978-3-032-02813-6","langid":"english","url":"https://www.wi2.uni-trier.de/shared/publications/Lenz2026NumbersDontLie.pdf","bibtex":"@inproceedings{Lenz2026NumbersDontLie,\n title = {Numbers {{Don}}'t {{Lie}}: {{Hybrid Extraction}} and~{{Validation}} of~{{Quantitative Statements}} in~{{Arguments}} with~{{Semi-structured Information}}},\n shorttitle = {Numbers {{Don}}'t {{Lie}}},\n booktitle = {{{KI}} 2025: {{Advances}} in {{Artificial Intelligence}}},\n author = {Lenz, Mirko and Dumani, Lorik and Schenkel, Ralf and Bergmann, Ralph},\n editor = {Braun, Tanya and Paa{\\ss}en, Benjamin and Stolzenburg, Frieder},\n year = {2026},\n series = {Lecture {{Notes}} in {{Computer Science}}},\n volume = {15956},\n pages = {77--90},\n publisher = {Springer Nature Switzerland},\n address = {Cham},\n doi = {10.1007/978-3-032-02813-6_6},\n abstract = {Evidence in arguments may be stated in various forms, including quantitative statements (i.e., numerical relations between entities). This measurable information can be validated against reliable sources like Wikipedia to combat the spread of misinformation. In this paper, we propose a four-step pipeline that combines rule-based techniques with prompting strategies for generative language models in a hybrid fashion. We use regular expressions to identify candidates in claim-premise structures, extract statements using GPT-4o, augment the data with tables from Wikipedia, and validate statements through retrieval-augmented generation (RAG). The pipeline is evaluated on two existing argumentation corpora and the generated dataset is manually annotated to assess the quality of our predictions, showing promising results for extraction and mixed results for validation. Our code and data are available to foster further research in this area.},\n isbn = {978-3-032-02813-6},\n langid = {english},\n url = {https://www.wi2.uni-trier.de/shared/publications/Lenz2026NumbersDontLie.pdf}\n}\n\n","author_short":["Lenz, M.","Dumani, L.","Schenkel, R.","Bergmann, R."],"editor_short":["Braun, T.","Paaßen, B.","Stolzenburg, F."],"key":"Lenz2026NumbersDontLie","id":"Lenz2026NumbersDontLie","bibbaseid":"lenz-dumani-schenkel-bergmann-numbersdontliehybridextractionandvalidationofquantitativestatementsinargumentswithsemistructuredinformation-2026","role":"author","urls":{"Paper":"https://www.wi2.uni-trier.de/shared/publications/Lenz2026NumbersDontLie.pdf"},"metadata":{"authorlinks":{}},"downloads":2},"bibtype":"inproceedings","biburl":"https://web.wi2.uni-trier.de/publications/WI2Publikationen.bib","dataSources":["MSp3DzP4ToPojqkFy"],"keywords":[],"search_terms":["numbers","don","lie","hybrid","extraction","validation","quantitative","statements","arguments","semi","structured","information","lenz","dumani","schenkel","bergmann"],"title":"Numbers Don't Lie: Hybrid Extraction and Validation of Quantitative Statements in Arguments with Semi-structured Information","year":2026,"downloads":2}