Idanwo ti o sopọ mọ itọsọna · Alabọde Ipele

Ibasepo isediwon lati Text adanwo

Test how relation extraction turns text into structured facts, and how its common pipelines and supervision methods can fail.

Jẹmọ itọsọna awọn ọnaIbasepo isediwon Lati Text
Ibeere 1 ti 8

Kini abajade aṣoju ti eto isediwon ibatan?

Awotẹlẹ gbogbo ibeere8 ibeere

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

  1. Kini abajade aṣoju ti eto isediwon ibatan?
    • Meta ti eleto ti o so awọn nkan meji pọ nipasẹ ibatan kan
    • A Gírámọ parse igi ti awọn gbolohun ọrọ
    • Dimegilio ikunsinu lati odi si rere
    • Itumọ gbolohun naa si ede miiran
  2. In a common pipeline for relation extraction, what may identify entity mentions before pair classification?
    • Ọrọ-si-ọrọ kolaginni
    • Ti a npè ni idanimọ nkankan
    • Pipa Pipa
    • Audio source separation
  3. Why can distant supervision produce noisy relation-extraction labels?
    • Knowledge bases contain no structured facts
    • It requires every sentence to be hand-labeled
    • A sentence may mention two linked entities without stating their knowledge-base relation
    • It removes entity mentions before training
  4. Bawo ni ọpọlọpọ awọn awoṣe RE ti o da lori ẹrọ oluyipada ṣe tọkasi iru orisii nkankan lati ṣe lẹtọ?
    • Nipa titumọ gbolohun naa ni akọkọ
    • Nipa piparẹ ọkan ninu awọn ile-iṣẹ naa
    • Nipa yiyipada ọrọ pada si ohun
    • Nipa fifi awọn ami ami ami pataki sii ni ayika awọn nkan meji
  5. Kini o ṣe iyatọ si isediwon ibatan 'ṣii' lati boṣewa RE?
    • O yọkuro awọn gbolohun ibatan taara lati inu ọrọ laisi eto ti a ti yan tẹlẹ ti awọn iru ibatan
    • O ṣiṣẹ nikan lori awọn nọmba
    • O nilo data aami diẹ sii ju eyikeyi ọna miiran lọ
    • O le rii ibatan kan nikan fun iwe-ipamọ
  6. Why include a no-relation label when classifying candidate entity pairs?
    • To translate the sentence into another language
    • To represent pairs for which the text does not assert a target relation
    • To replace named entity recognition
    • To guarantee that every pair has a relation
  7. What does document-level relation extraction allow a system to use?
    • Only the spelling of a single entity
    • Audio features from a spoken recording
    • Evidence that may be spread across multiple sentences
    • A fixed relation without reading text
  8. How does open relation extraction differ from a closed relation schema?
    • It cannot use text as input
    • It always requires a knowledge base to label every sentence
    • It only identifies entity boundaries
    • It can express relation phrases without choosing only from a predefined label inventory