{
  "$schema": "https://posters.science/schema/v0.2/poster_schema.json",
  "titles": [
    {
      "title": "A Dataset for Metaphor Detection in Early Medieval Hebrew Poetry"
    }
  ],
  "descriptions": [
    {
      "description": "The corpus of late antique and medieval Hebrew is vast and represents a crucial linguistic and cultural bridge between Biblical and modern Hebrew. Poetry is prominent in texts from these time periods and one of its characteristics is the use of metaphor. Distinguishing figurative and literal words or phrases is a major task for scholars in the Humanities, especially in the fields of literature, linguistics, and hermeneutics. We present a new, challenging dataset of early medieval Hebrew poetry with expert annotations of metaphors with baseline results.",
      "descriptionType": "Other"
    }
  ],
  "dates": [
    {
      "date": "2026-06-05",
      "dateType": "Submitted",
      "dateInformation": "Submitted to Zenodo through Posters.science"
    }
  ],
  "types": {
    "resourceType": "Conference Poster",
    "resourceTypeGeneral": "Poster"
  },
  "identifiers": [
    {
      "identifier": "10.5281/zenodo.509690",
      "identifierType": "DOI"
    }
  ],
  "creators": [
    {
      "name": "Adir Cohen",
      "givenName": "Adir",
      "familyName": "Cohen",
      "nameType": "Personal",
      "nameIdentifiers": [
        {
          "nameIdentifier": "https://orcid.org/0009-0009-5055-4128",
          "nameIdentifierScheme": "ORCID",
          "schemeURI": "https://orcid.org"
        }
      ],
      "affiliation": [
        {
          "name": "Technion – Israel Institute of Technology",
          "schemeURI": "https://ror.org",
          "affiliationIdentifier": "https://ror.org/03qryx823",
          "affiliationIdentifierScheme": "ROR"
        },
        {
          "name": "Open University of Israel",
          "schemeURI": "https://ror.org",
          "affiliationIdentifier": "https://ror.org/027z64205",
          "affiliationIdentifierScheme": "ROR"
        }
      ]
    },
    {
      "name": "Ophir Munz",
      "givenName": "Ophir",
      "familyName": "Munz",
      "nameType": "Personal",
      "affiliation": [
        {
          "name": "Technion – Israel Institute of Technology",
          "schemeURI": "https://ror.org",
          "affiliationIdentifier": "https://ror.org/03qryx823",
          "affiliationIdentifierScheme": "ROR"
        }
      ]
    },
    {
      "name": "Oren Mishali",
      "givenName": "Oren",
      "familyName": "Mishali",
      "nameType": "Personal",
      "nameIdentifiers": [
        {
          "nameIdentifier": "https://orcid.org/0009-0000-3305-5292",
          "nameIdentifierScheme": "ORCID",
          "schemeURI": "https://orcid.org"
        }
      ],
      "affiliation": [
        {
          "name": "Technion – Israel Institute of Technology",
          "schemeURI": "https://ror.org",
          "affiliationIdentifier": "https://ror.org/03qryx823",
          "affiliationIdentifierScheme": "ROR"
        }
      ]
    },
    {
      "name": "Michael Toker",
      "givenName": "Michael",
      "familyName": "Toker",
      "nameType": "Personal",
      "nameIdentifiers": [
        {
          "nameIdentifier": "https://orcid.org/0009-0002-9293-868X",
          "nameIdentifierScheme": "ORCID",
          "schemeURI": "https://orcid.org"
        }
      ],
      "affiliation": [
        {
          "name": "Technion – Israel Institute of Technology",
          "schemeURI": "https://ror.org",
          "affiliationIdentifier": "https://ror.org/03qryx823",
          "affiliationIdentifierScheme": "ROR"
        }
      ]
    },
    {
      "name": "Benny Kimelfeld",
      "givenName": "Benny",
      "familyName": "Kimelfeld",
      "nameType": "Personal",
      "affiliation": [
        {
          "name": "Technion – Israel Institute of Technology",
          "schemeURI": "https://ror.org",
          "affiliationIdentifier": "https://ror.org/03qryx823",
          "affiliationIdentifierScheme": "ROR"
        }
      ]
    },
    {
      "name": "Yonatan Belinkov",
      "givenName": "Yonatan",
      "familyName": "Belinkov",
      "nameType": "Personal",
      "affiliation": [
        {
          "name": "Technion – Israel Institute of Technology",
          "schemeURI": "https://ror.org",
          "affiliationIdentifier": "https://ror.org/03qryx823",
          "affiliationIdentifierScheme": "ROR"
        }
      ]
    }
  ],
  "publisher": {
    "name": "Zenodo",
    "publisherIdentifier": "https://doi.org/10.17616/R3QP53",
    "publisherIdentifierScheme": "DOI",
    "schemeURI": "https://doi.org"
  },
  "publicationYear": 2026,
  "subjects": [
    {
      "subject": "metaphor detection"
    },
    {
      "subject": "early medieval Hebrew poetry"
    },
    {
      "subject": "natural language processing"
    }
  ],
  "language": "en",
  "relatedIdentifiers": [
    {
      "relationType": "IsDescribedBy",
      "relatedIdentifier": "https://posters.science/discover/26713",
      "resourceTypeGeneral": "Other",
      "relatedIdentifierType": "URL"
    }
  ],
  "formats": [
    "application/pdf"
  ],
  "version": "1",
  "rightsList": [
    {
      "rights": "Creative Commons Zero v1.0 Universal",
      "rightsUri": "https://creativecommons.org/publicdomain/zero/1.0/legalcode",
      "rightsIdentifier": "CC0-1.0",
      "rightsIdentifierScheme": "SPDX",
      "schemeUri": "https://spdx.org/licenses/"
    }
  ],
  "fundingReferences": [
    {
      "awardUri": "https://google.com",
      "funderName": "test"
    }
  ],
  "conference": {
    "conferenceName": "dfas",
    "conferenceYear": 2026
  },
  "content": {
    "sections": [
      {
        "sectionTitle": "Introduction",
        "sectionContent": "The corpus of late antique and medieval Hebrew is vast and represents a crucial linguistic and cultural bridge between Biblical and modern Hebrew. Poetry is prominent in texts from these time periods and one of its characteristics is the use of metaphor. Distinguishing figurative and literal words or phrases is a major task for scholars in the Humanities, especially in the fields of literature, linguistics, and hermeneutics. We present a new, challenging dataset of early medieval Hebrew poetry with expert annotations of metaphors with baseline results."
      },
      {
        "sectionTitle": "The Dataset",
        "sectionContent": "• Early medieval Hebrew poems. • 180 labeled texts, 7,000 sentences, 45,000 words. • 15% of the words are metaphors."
      },
      {
        "sectionTitle": "Inter Annotator Agreement",
        "sectionContent": "Inter-annotator agreement evaluates how consistently different annotators label data for NLP research (measured in Cohen's kappa). The score for our dataset is 0.618 (considered substantial)."
      },
      {
        "sectionTitle": "Methods Fine-tuning for metaphor detection",
        "sectionContent": "Fine-tuning for metaphor detection"
      },
      {
        "sectionTitle": "LM adaptation",
        "sectionContent": "Pre-trained Language Model"
      },
      {
        "sectionTitle": "Language models (pre-trained & adapted)",
        "sectionContent": "Data Size (words) Data Type\nBEREL 220 Milion Medieval\nAlephBERT\nAlephBERT Adapted 2 Bilion\n2 Bilion + 10 Milion Modern\nModern + Medieval\nPre-trained/Adapted Language Model"
      },
      {
        "sectionTitle": "Tokenizer",
        "sectionContent": "ונידי חלשמב ןובציע"
      },
      {
        "sectionTitle": "Results",
        "sectionContent": "• AlephBERT is outperformed by AlephBERT Adapted. • BEREL, pre-trained on medieval texts, shows the best results. f1 Score on test set 48.3\nBEREL 49 36.75 24.5 12.25 0 44.404\nAlephBERT _ידי <mask> ןוב_ <mask>\n_ידי חלשמב ןוב_ _ציע"
      },
      {
        "sectionTitle": "Tokenizer",
        "sectionContent": "ונידי חלשמב ןובציע"
      },
      {
        "sectionTitle": "Error Analysis",
        "sectionContent": "ויתקשח חוחינל דלי (A child for a scent - I have desired)\nBerel\nויתקשח חוחינל דלי\nExpert explanation: The child is a non-metaphorical epithet for Isaac. 'Scent' is a metonymic term for 'sacrifice'."
      },
      {
        "sectionContent": "Adir Cohen adircohen@campus.technion.ac.il Ophir M unz ·· - Manor 2 ophirmm@openu.ac.il 1\nOren Mishali omishali@cs.technion.ac.il 1\nMichael Toker tok@cs.technion.ac.il 1\nBenny Kimelfeld bennyk@cs.technion.ac.il 1\nYonatan Belinkov belinkov@technion.ac.il\nTechnion − Israel Institute of Technology"
      },
      {
        "sectionContent": "‫םי‬ ‫יפוטשכ‬ ‫,תורזגב‬ ‫ונעבט‬\n‫רנ‬ ‫רעביו‬ ‫המשנ‬ ‫תיצה‬\n‫ירפ‬ ‫ושע‬ ‫אל‬ ‫ירישע‬ ‫רוד‬ ‫דע‬ English\nWe drowned in decrees, like sea washed\nIgnite a soul, fire a candle\nUntil the tenth generation they did not bear fruit Meaning\nThere are too many decrees\nCreate a soul\nFor ten generations, they did no good deeds\nInter Annotator Agreement"
      },
      {
        "sectionContent": "detection LM adaptation\n‫חלשמב‬\nPre-trained Language Model ‫_ציע‬ A medieval Hebrew poem from the Worms Prayerbook, Germany, 13th Century, The National Library of Israel, \"Ktiv\" Project\nLanguage models (pre-trained & adapted)"
      },
      {
        "sectionContent": "‫ונ_‬ ‫_ידי‬ ‫חלשמב‬ ‫ןוב_‬ ‫_ציע‬\nTokenizer\n‫ונידי‬ ‫חלשמב‬ ‫ןובציע‬\nResults"
      },
      {
        "sectionContent": "‫_ידי‬ ‫חלשמב‬ ‫ןוב_‬ ‫_ציע‬\nTokenizer\n‫ונידי‬ ‫חלשמב‬ ‫ןובציע‬\n46.56"
      },
      {
        "sectionContent": "‫ונ_‬ Trained on additional prose and poetic medieval texts (Project Ben-Yehuda*) + Piyyut dataset\n*Project Ben-Yehuda is a volunteer-based online library of free Hebrew literature.\nError Analysis"
      },
      {
        "sectionContent": "‫ויתקשח‬ ‫חוחינל‬ ‫דלי‬"
      }
    ]
  },
  "researchField": "Humanities"
}