Abstract
We propose a new end-to-end method for extending a Knowledge Graph (KG) from tables. Existing techniques tend to interpret tables by focusing on information that is already in the KG, and therefore tend to extract many redundant facts. Our method aims to find more novel facts. We introduce a new technique for table interpretation based on a scalable graphical model using entity similarities. Our method further disambiguates cell values using KG embeddings as additional ranking method. Other distinctive features are the lack of assumptions about the underlying KG and the enabling of a fine-grained tuning of the precision/recall trade-off of extracted facts. Our experiments show that our approach has a higher recall during the interpretation process than the state-of-the-art, and is more resistant against the bias observed in extracting mostly redundant facts since it produces more novel extractions.
| Original language | English |
|---|---|
| Title of host publication | The Semantic Web – ISWC 2019 |
| Subtitle of host publication | 18th International Semantic Web Conference, Auckland, New Zealand, October 26–30, 2019, Proceedings |
| Editors | Chiara Ghidini, Olaf Hartig, Maria Maleshkova, Vojtech Svátek, Isabel Cruz, Aidan Hogan, Jie Song, Maxime Lefrançois, Fabien Gandon |
| Publisher | Springer |
| Pages | 364-381 |
| Number of pages | 18 |
| Volume | 1 |
| ISBN (Electronic) | 9783030307936 |
| ISBN (Print) | 9783030307929 |
| DOIs | |
| Publication status | Published - 2019 |
| Event | 18th International Semantic Web Conference, ISWC 2019 - Auckland, New Zealand Duration: 26 Oct 2019 → 30 Oct 2019 |
Publication series
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Volume | 11778 LNCS |
| ISSN (Print) | 0302-9743 |
| ISSN (Electronic) | 1611-3349 |
Conference
| Conference | 18th International Semantic Web Conference, ISWC 2019 |
|---|---|
| Country | New Zealand |
| City | Auckland |
| Period | 26/10/19 → 30/10/19 |
Fingerprint
Cite this
}
Extracting Novel Facts from Tables for Knowledge Graph Completion. / Kruit, Benno; Boncz, Peter; Urbani, Jacopo.
The Semantic Web – ISWC 2019: 18th International Semantic Web Conference, Auckland, New Zealand, October 26–30, 2019, Proceedings. ed. / Chiara Ghidini; Olaf Hartig; Maria Maleshkova; Vojtech Svátek; Isabel Cruz; Aidan Hogan; Jie Song; Maxime Lefrançois; Fabien Gandon. Vol. 1 Springer, 2019. p. 364-381 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11778 LNCS).Research output: Chapter in Book / Report / Conference proceeding › Conference contribution › Academic › peer-review
TY - GEN
T1 - Extracting Novel Facts from Tables for Knowledge Graph Completion
AU - Kruit, Benno
AU - Boncz, Peter
AU - Urbani, Jacopo
PY - 2019
Y1 - 2019
N2 - We propose a new end-to-end method for extending a Knowledge Graph (KG) from tables. Existing techniques tend to interpret tables by focusing on information that is already in the KG, and therefore tend to extract many redundant facts. Our method aims to find more novel facts. We introduce a new technique for table interpretation based on a scalable graphical model using entity similarities. Our method further disambiguates cell values using KG embeddings as additional ranking method. Other distinctive features are the lack of assumptions about the underlying KG and the enabling of a fine-grained tuning of the precision/recall trade-off of extracted facts. Our experiments show that our approach has a higher recall during the interpretation process than the state-of-the-art, and is more resistant against the bias observed in extracting mostly redundant facts since it produces more novel extractions.
AB - We propose a new end-to-end method for extending a Knowledge Graph (KG) from tables. Existing techniques tend to interpret tables by focusing on information that is already in the KG, and therefore tend to extract many redundant facts. Our method aims to find more novel facts. We introduce a new technique for table interpretation based on a scalable graphical model using entity similarities. Our method further disambiguates cell values using KG embeddings as additional ranking method. Other distinctive features are the lack of assumptions about the underlying KG and the enabling of a fine-grained tuning of the precision/recall trade-off of extracted facts. Our experiments show that our approach has a higher recall during the interpretation process than the state-of-the-art, and is more resistant against the bias observed in extracting mostly redundant facts since it produces more novel extractions.
UR - http://www.scopus.com/inward/record.url?scp=85075720339&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85075720339&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-30793-6_21
DO - 10.1007/978-3-030-30793-6_21
M3 - Conference contribution
SN - 9783030307929
VL - 1
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 364
EP - 381
BT - The Semantic Web – ISWC 2019
A2 - Ghidini, Chiara
A2 - Hartig, Olaf
A2 - Maleshkova, Maria
A2 - Svátek, Vojtech
A2 - Cruz, Isabel
A2 - Hogan, Aidan
A2 - Song, Jie
A2 - Lefrançois, Maxime
A2 - Gandon, Fabien
PB - Springer
ER -