{
 "uuid": "6d0b7cc6-36da-4ee0-b0b4-60da175d4dbd",
 "handle": "uniba/95827",
 "type": "Publication",
 "name": "Emotion-Conditioned Text Generation through Automatic Prompt Optimization",
 "url": "/entities/publication/6d0b7cc6-36da-4ee0-b0b4-60da175d4dbd/",
 "lastModified": "2024-10-15T09:15:25.171+00:00",
 "metadata": {
  "cris.virtual.author-orcid": [
   {
    "value": null
   },
   {
    "value": "0000-0002-2014-6619"
   }
  ],
  "cris.virtualsource.author-orcid": [
   {
    "value": "d98d9cbc-8f02-46bf-9e84-74797e598233"
   },
   {
    "value": "a2a36d65-e914-48fb-8e87-a2796ce66cb3"
   }
  ],
  "dc.contributor.author": [
   {
    "value": "Menchaca Resendiz, Yarik",
    "authority": "d98d9cbc-8f02-46bf-9e84-74797e598233"
   },
   {
    "value": "Klinger, Roman",
    "authority": "a2a36d65-e914-48fb-8e87-a2796ce66cb3"
   }
  ],
  "dc.date.accessioned": [
   {
    "value": "2024-06-20T06:26:25Z"
   }
  ],
  "dc.date.available": [
   {
    "value": "2024-06-20T06:26:25Z"
   }
  ],
  "dc.date.issued": [
   {
    "value": "2024"
   }
  ],
  "dc.description.abstract": [
   {
    "value": "Conditional natural language generation methods often require either expensive fine-tuning or training a large language model from scratch. Both are unlikely to lead to good results without a substantial amount of data and computational resources. Prompt learning without changing the parameters of a large language model presents a promising alternative. It is a cost-effective approach, while still achieving competitive results. While this procedure is now established for zero- and few-shot text classification and structured prediction, it has received limited attention in conditional text generation. We present the first automatic prompt optimization approach for emotion-conditioned text generation with instruction-fine-tuned models. Our method uses an iterative optimization procedure that changes the prompt by adding, removing, or replacing tokens. As objective function, we only require a text classifier that measures the realization of the conditional variable in the generated text. We evaluate the method on emotion-conditioned text generation with a focus on event reports and compare it to manually designed prompts that also act as the seed for the optimization procedure. The optimized prompts achieve 0.75 macro-average F1 to fulfill the emotion condition in contrast to manually designed seed prompts with only 0.22 macro-average F1."
   }
  ],
  "dc.identifier.uri": [
   {
    "value": "https://fis.uni-bamberg.de/handle/uniba/95827"
   }
  ],
  "dc.identifier.urn": [
   {
    "value": "urn:nbn:de:bvb:473-irb-958278"
   }
  ],
  "dc.language.iso": [
   {
    "value": "eng"
   }
  ],
  "dc.publisher": [
   {
    "value": "Otto-Friedrich-Universität"
   }
  ],
  "dc.publisher.place": [
   {
    "value": "Bamberg"
   }
  ],
  "dc.relation.authororgunit": [
   {
    "value": null,
    "authority": "#PLACEHOLDER_PARENT_METADATA_VALUE#"
   },
   {
    "value": "Fundamentals of Natural Language Processing",
    "authority": "e42c6d56-e293-45c9-9582-a71b5e4aee22"
   }
  ],
  "dc.subject": [
   {
    "value": "Emotion-Conditioned Text Generation"
   }
  ],
  "dc.subject.ddc": [
   {
    "value": "004",
    "authority": "ddc:004"
   }
  ],
  "dc.title": [
   {
    "value": "Emotion-Conditioned Text Generation through Automatic Prompt Optimization"
   }
  ],
  "dc.type": [
   {
    "value": "conferenceobject"
   }
  ],
  "dspace.entity.type": [
   {
    "value": "Publication"
   }
  ],
  "ubg.accessRights.copyright": [
   {
    "value": "true"
   }
  ],
  "ubg.accessRights.licence": [
   {
    "value": "CC BY 4.0"
   }
  ],
  "ubg.date.published": [
   {
    "value": "2023"
   }
  ],
  "ubg.fulltexticon": [
   {
    "value": "true"
   }
  ],
  "ubg.generated.csl": [
   {
    "value": "<div class=\"csl-bib-body\">\n  <div class=\"csl-entry\">Menchaca Resendiz, Yarik; Klinger, Roman (2024): Emotion-Conditioned Text Generation through Automatic Prompt Optimization, in: Bamberg: Otto-Friedrich-Universität, S. 24–30.</div>\n</div>"
   }
  ],
  "ubg.intern.org": [
   {
    "value": "Fundamentals of Natural Language Processing",
    "authority": "e42c6d56-e293-45c9-9582-a71b5e4aee22"
   }
  ],
  "ubg.intern.source": [
   {
    "value": "Proceedings of the 1st Workshop on Taming Large Language Models : Controllability in the era of Interactive Assistants! / Devamanyu Hazarika, Xiangru Robert Tang, Di Jin (Hg.). - Prag : Association for Computational Linguistics, 2023, S. 24–30."
   }
  ],
  "ubg.intern.vonfeld": [
   {
    "value": "",
    "language": "de_DE"
   }
  ],
  "ubg.intern.vonfeldeditor": [
   {
    "value": "",
    "language": "de_DE"
   }
  ],
  "ubg.internatorientation": [
   {
    "value": "Ja"
   }
  ],
  "ubg.openaccess": [
   {
    "value": "Ja"
   }
  ],
  "ubg.pages.range": [
   {
    "value": "24-30"
   }
  ],
  "ubg.peerreview": [
   {
    "value": "Ja"
   }
  ],
  "ubg.relation.project": [
   {
    "value": "Interactive Prompt Optimization with the Human in the Loop for Natural Language Understanding Model Development and Intervention",
    "authority": "098980c8-a33f-4560-8f62-0ebe368d248f"
   }
  ],
  "ubg.subject.gndsw": [
   {
    "value": "Maschinelles Lernen"
   },
   {
    "value": "Generierung <Sprache>"
   },
   {
    "value": "Gefühl"
   },
   {
    "value": "Prompt Engineering"
   }
  ],
  "ubg.subject.rvk": [
   {
    "value": "ST 306"
   }
  ],
  "ubg.submission.type": [
   {
    "value": "full-text"
   }
  ],
  "ubg.thesis.grantor": [
   {
    "value": "Uni"
   }
  ],
  "ubg.version.type": [
   {
    "value": "publishedVersion"
   }
  ],
  "ubg.version.versionof": [
   {
    "value": "uniba/93873",
    "authority": "01ab7d84-8363-4447-9da7-4616b0c46f9b"
   }
  ]
 },
 "relations": {},
 "files": [
  {
   "uuid": "6da435f7-93cf-47c4-ace5-b62416d6a66e",
   "bundle": "ORIGINAL",
   "name": "fisba95827.pdf",
   "size": 1119899,
   "md5": "0ce9c5215afde402f1b5035ff1feda52",
   "mimetype": "application/pdf",
   "url": "/files/6da435f7-93cf-47c4-ace5-b62416d6a66e/fisba95827.pdf",
   "description": null
  }
 ]
}