Tweet
Tweet

Slide 1

Slide 1 text

Neural Relation Extraction for Knowledge Base Enrichment Bayu Distiawan Trisedya, Gerhard Weikum, Jianzhong Qi, Rui Zhang ACL 2019 ঺հऀ: দాߞ࢙ (ཧݚAIP/౦๺େ) ୈ11ճ࠷ઌ୺ NLP ษڧձ (γϣʔτ࿮) εϥΠυதͷਤද͸ಛʹ஫ऍ͕ͳ͍ݶΓ࿦จ͔ΒͷҾ༻Ͱ͢ 1 YAGOͷ։ൃऀ

Slide 2

Slide 2 text

  • MEKG'"$) >9 GD  – KG Enrichment • =1 (: Seq2Seq    2 • @C – ?F#-*>986 IB!0' '34L< – K7IB JA;",+&,  n-gram Attention – ?F#-*5).+/*/ %  Precision:H 2

Slide 3

Slide 3 text

: (Wikidata) 3

Slide 4

Slide 4 text

KG Enrichment  Information Extraction ΤϯςΟςΟͱؔ܎Λ WikidataͷIDʹਖ਼نԽ͍ͨ͠ NED => ؔ܎நग़ ͳύΠϓϥΠϯ͕ϕʔεϥΠϯ 4

Slide 5

Slide 5 text

 :      New York University is a private university in Manhattan. Encoder (LSTM) N-gram Attention Decoder (LSTM) Triple Classifier ޻෉ᶄ ޻෉ᶅ Joint Embedding ޻෉ᶃ ೖྗจ ग़ྗ 5 ೖग़ྗͷϖΞ͸ Distant SupervisionͰੜ੒

Slide 6

Slide 6 text

: KG      • TransE + Skip-gram –  Anchor Context Model [Yamada 16] 6 New York University is a private university in Manhattan Q49210 is a Q902104 in Q11299 J = +

Slide 7

Slide 7 text

: N-gram Attention •   Entity-mention    – ”New York University” • “New York”, “York University” … 7 3-gram·Ͱͷn-gram

Slide 8

Slide 8 text

: Triple Classifier • 5<"+(/   5< (-% – 9: • TransE  '- 04&,)-= 7@23$#*!&-   6;.> – (h + r – t) – .>: Negative Sample 81? 8

Slide 9

Slide 9 text

 • ;87/! Distant Supervision – Co-reference  – -4#(6#' 3 2  • ;8!* )  ., &!'1 – $ ;8!9  • )  .,0: "% +  (533.5) 9

Slide 10

Slide 10 text

 10 ύΠϓϥΠ ϯ͸μϝ N-gram Attention͸ ޮՌൃش ClassifierͰ Precisionվળ

Slide 11

Slide 11 text

 New York University is a private university in Manhattan. Encoder (LSTM) N-gram Attention Decoder (LSTM) Triple Classifier ޻෉ᶄ ޻෉ᶅ Joint Embedding ޻෉ᶃ ೖྗจ ग़ྗ 11 ೖग़ྗͷϖΞ͸ Distant SupervisionͰੜ੒

Slide 12

Slide 12 text

 • ! •   – DS   train/test    test  12