web_nlg

  • विवरण :

डेटा में (डीबीपीडिया) [ https://wiki.dbpedia.org/ ] से निकाले गए फॉर्म सब्जेक्ट-प्रेडिकेट-ऑब्जेक्ट के 1 से 7 ट्रिपल के सेट होते हैं और प्राकृतिक भाषा का पाठ जो इन ट्रिपल्स का मौखिककरण है। परीक्षण डेटा 15 अलग-अलग डोमेन तक फैला हुआ है जहाँ प्रशिक्षण डेटा में केवल 10 दिखाई देते हैं। डेटासेट एक मानकीकृत तालिका प्रारूप का अनुसरण करता है।

विभाजित करना उदाहरण
'test_all' 4,928
'test_unseen' 2,433
'train' 18,102
'validation' 2,268
  • फ़ीचर संरचना :
FeaturesDict({
    'input_text': FeaturesDict({
        'context': string,
        'table': Sequence({
            'column_header': string,
            'content': string,
            'row_number': int16,
        }),
    }),
    'target_text': string,
})
  • फ़ीचर दस्तावेज़ीकरण :
विशेषता कक्षा आकार डीटाइप विवरण
विशेषताएं डिक्ट
इनपुट टेक्स्ट विशेषताएं डिक्ट
input_text/context टेन्सर डोरी
इनपुट_टेक्स्ट/टेबल क्रम
input_text/टेबल/column_header टेन्सर डोरी
input_text/तालिका/सामग्री टेन्सर डोरी
इनपुट_टेक्स्ट/टेबल/पंक्ति_नंबर टेन्सर int16
target_text टेन्सर डोरी
  • उद्धरण :
@inproceedings{gardent2017creating,
    title = ""Creating Training Corpora for {NLG} Micro-Planners"",
    author = ""Gardent, Claire  and
      Shimorina, Anastasia  and
      Narayan, Shashi  and
      Perez-Beltrachini, Laura"",
    booktitle = ""Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)"",
    month = jul,
    year = ""2017"",
    address = ""Vancouver, Canada"",
    publisher = ""Association for Computational Linguistics"",
    doi = ""10.18653/v1/P17-1017"",
    pages = ""179--188"",
    url = ""https://www.aclweb.org/anthology/P17-1017.pdf""
}