XINA: Explainable Instance Alignment using Dominance Relationship

Jinyoung Yeo, Haeju Park, Sanghoon Lee, Eric Wonhee Lee, Seung Won Hwang

Research output: Contribution to journalArticle

Abstract

Over the past few years, knowledge bases (KBs) like DBPedia, Freebase, and YAGO have accumulated a massive amount of knowledge from web data. Despite their seemingly large size, however, individual KBs often lack comprehensive information on any given domain. For example, over 70% of people on Freebase lack information on place of birth. For this reason, the complementary nature across different KBs motivates their integration through a process of aligning instances. Meanwhile, since application-level machine systems, such as medical diagnosis, have heavily relied on KBs, it is necessary to provide users with trustworthy reasons why the alignment decisions are made. To address this problem, we propose a new paradigm, explainable instance alignment (XINA), which provides user-understandable explanations for alignment decisions. Specifically, given an alignment candidate, XINA replaces existing scalar representation of an aggregated score, by decision- and explanation-vector spaces for machine decision and user understanding, respectively. To validate XINA, we perform extensive experiments on real-world KBs and show that XINA achieves comparable performance with state-of-the-arts, even with far less human effort.

Original languageEnglish
JournalIEEE Transactions on Knowledge and Data Engineering
DOIs
Publication statusAccepted/In press - 2018 Jan 1

Fingerprint

Vector spaces
Experiments

All Science Journal Classification (ASJC) codes

  • Information Systems
  • Computer Science Applications
  • Computational Theory and Mathematics

Cite this

@article{505ace352b984332b5dc8e5e322e0256,
title = "XINA: Explainable Instance Alignment using Dominance Relationship",
abstract = "Over the past few years, knowledge bases (KBs) like DBPedia, Freebase, and YAGO have accumulated a massive amount of knowledge from web data. Despite their seemingly large size, however, individual KBs often lack comprehensive information on any given domain. For example, over 70{\%} of people on Freebase lack information on place of birth. For this reason, the complementary nature across different KBs motivates their integration through a process of aligning instances. Meanwhile, since application-level machine systems, such as medical diagnosis, have heavily relied on KBs, it is necessary to provide users with trustworthy reasons why the alignment decisions are made. To address this problem, we propose a new paradigm, explainable instance alignment (XINA), which provides user-understandable explanations for alignment decisions. Specifically, given an alignment candidate, XINA replaces existing scalar representation of an aggregated score, by decision- and explanation-vector spaces for machine decision and user understanding, respectively. To validate XINA, we perform extensive experiments on real-world KBs and show that XINA achieves comparable performance with state-of-the-arts, even with far less human effort.",
author = "Jinyoung Yeo and Haeju Park and Sanghoon Lee and Lee, {Eric Wonhee} and Hwang, {Seung Won}",
year = "2018",
month = "1",
day = "1",
doi = "10.1109/TKDE.2018.2881956",
language = "English",
journal = "IEEE Transactions on Knowledge and Data Engineering",
issn = "1041-4347",
publisher = "IEEE Computer Society",

}

XINA : Explainable Instance Alignment using Dominance Relationship. / Yeo, Jinyoung; Park, Haeju; Lee, Sanghoon; Lee, Eric Wonhee; Hwang, Seung Won.

In: IEEE Transactions on Knowledge and Data Engineering, 01.01.2018.

Research output: Contribution to journalArticle

TY - JOUR

T1 - XINA

T2 - Explainable Instance Alignment using Dominance Relationship

AU - Yeo, Jinyoung

AU - Park, Haeju

AU - Lee, Sanghoon

AU - Lee, Eric Wonhee

AU - Hwang, Seung Won

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Over the past few years, knowledge bases (KBs) like DBPedia, Freebase, and YAGO have accumulated a massive amount of knowledge from web data. Despite their seemingly large size, however, individual KBs often lack comprehensive information on any given domain. For example, over 70% of people on Freebase lack information on place of birth. For this reason, the complementary nature across different KBs motivates their integration through a process of aligning instances. Meanwhile, since application-level machine systems, such as medical diagnosis, have heavily relied on KBs, it is necessary to provide users with trustworthy reasons why the alignment decisions are made. To address this problem, we propose a new paradigm, explainable instance alignment (XINA), which provides user-understandable explanations for alignment decisions. Specifically, given an alignment candidate, XINA replaces existing scalar representation of an aggregated score, by decision- and explanation-vector spaces for machine decision and user understanding, respectively. To validate XINA, we perform extensive experiments on real-world KBs and show that XINA achieves comparable performance with state-of-the-arts, even with far less human effort.

AB - Over the past few years, knowledge bases (KBs) like DBPedia, Freebase, and YAGO have accumulated a massive amount of knowledge from web data. Despite their seemingly large size, however, individual KBs often lack comprehensive information on any given domain. For example, over 70% of people on Freebase lack information on place of birth. For this reason, the complementary nature across different KBs motivates their integration through a process of aligning instances. Meanwhile, since application-level machine systems, such as medical diagnosis, have heavily relied on KBs, it is necessary to provide users with trustworthy reasons why the alignment decisions are made. To address this problem, we propose a new paradigm, explainable instance alignment (XINA), which provides user-understandable explanations for alignment decisions. Specifically, given an alignment candidate, XINA replaces existing scalar representation of an aggregated score, by decision- and explanation-vector spaces for machine decision and user understanding, respectively. To validate XINA, we perform extensive experiments on real-world KBs and show that XINA achieves comparable performance with state-of-the-arts, even with far less human effort.

UR - http://www.scopus.com/inward/record.url?scp=85056692858&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85056692858&partnerID=8YFLogxK

U2 - 10.1109/TKDE.2018.2881956

DO - 10.1109/TKDE.2018.2881956

M3 - Article

AN - SCOPUS:85056692858

JO - IEEE Transactions on Knowledge and Data Engineering

JF - IEEE Transactions on Knowledge and Data Engineering

SN - 1041-4347

ER -