國立高雄大學圖資館 |

Language: English

Back

Toward Deep Language Understanding: ...

Orfan, Jansen.

Toward Deep Language Understanding: Methods for Learning Conceptual Knowledge from Definitions.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Toward Deep Language Understanding: Methods for Learning Conceptual Knowledge from Definitions.
Author:	Orfan, Jansen.
Published:	Ann Arbor : ProQuest Dissertations & Theses, 2020
Description:	183 p.
Notes:	Source: Dissertations Abstracts International, Volume: 82-05, Section: B.
Notes:	Advisor: Allen, James F.
Contained By:	Dissertations Abstracts International82-05B.
Subject:	Computer science.
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28092953
ISBN:	9798678180292

Toward Deep Language Understanding: Methods for Learning Conceptual Knowledge from Definitions.
Orfan, Jansen.

Toward Deep Language Understanding: Methods for Learning Conceptual Knowledge from Definitions. - Ann Arbor : ProQuest Dissertations & Theses, 2020 - 183 p.

Source: Dissertations Abstracts International, Volume: 82-05, Section: B.

Thesis (Ph.D.)--University of Rochester, 2020.

This item must not be sold to any third party vendors.

Knowledge based AI is a fundamental paradigm that has made continued progress since early expert systems in the 1970s. While inference techniques may be developed to perform a certain task in theory, acquiring the knowledge to perform it well in practice is another matter. For limited tasks, knowledge could be added manually or acquired using existing knowledge extraction techniques like pattern matching. However, as the scope of a task becomes more general the complexity and amount of knowledge becomes too great to rely on current approaches. This knowledge acquisition bottleneck greatly limits any inference technique and is a common obstacle in further developing very general knowledge based AI like common sense reasoning and \lu{}. In this thesis we will examine the knowledge bottle neck in \lu{} and how it relates to typical reasoning techniques in many \lu{} tasks. We examine the current state of knowledge available for those tasks as well as the types and scale of knowledge that is lacking. We then discuss techniques for acquiring symbolic knowledge about general concepts from dictionary sources using semantic parsing. We propose ways of augmenting the acquired knowledge with subtext that tends to be ignored by current methods which interpret definitions literally. This allows us to abstract concepts into frames and discover semantic relationships between concepts. We show that our process of adding subtext acquires more semantic relationships of similar accuracy compared to methods without added subtext.

ISBN: 9798678180292Subjects--Topical Terms:

199325
Computer science.
Subjects--Index Terms:

Knowledge based AI

Toward Deep Language Understanding: Methods for Learning Conceptual Knowledge from Definitions.
LDR:02684nmm a2200349 4500 001 594622
005 20210521101710.5
008 210917s2020 ||||||||||||||||| ||eng d
020 $a 9798678180292
035 $a (MiAaPQ)AAI28092953
035 $a AAI28092953
040 $a MiAaPQ $c MiAaPQ
100 1 $a Orfan, Jansen. $3 886666
245 1 0 $a Toward Deep Language Understanding: Methods for Learning Conceptual Knowledge from Definitions.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2020
300 $a 183 p.
500 $a Source: Dissertations Abstracts International, Volume: 82-05, Section: B.
500 $a Advisor: Allen, James F.
502 $a Thesis (Ph.D.)--University of Rochester, 2020.
506 $a This item must not be sold to any third party vendors.
520 $a Knowledge based AI is a fundamental paradigm that has made continued progress since early expert systems in the 1970s. While inference techniques may be developed to perform a certain task in theory, acquiring the knowledge to perform it well in practice is another matter. For limited tasks, knowledge could be added manually or acquired using existing knowledge extraction techniques like pattern matching. However, as the scope of a task becomes more general the complexity and amount of knowledge becomes too great to rely on current approaches. This knowledge acquisition bottleneck greatly limits any inference technique and is a common obstacle in further developing very general knowledge based AI like common sense reasoning and \lu{}. In this thesis we will examine the knowledge bottle neck in \lu{} and how it relates to typical reasoning techniques in many \lu{} tasks. We examine the current state of knowledge available for those tasks as well as the types and scale of knowledge that is lacking. We then discuss techniques for acquiring symbolic knowledge about general concepts from dictionary sources using semantic parsing. We propose ways of augmenting the acquired knowledge with subtext that tends to be ignored by current methods which interpret definitions literally. This allows us to abstract concepts into frames and discover semantic relationships between concepts. We show that our process of adding subtext acquires more semantic relationships of similar accuracy compared to methods without added subtext.
590 $a School code: 0188.
650 4 $a Computer science. $3 199325
650 4 $a Artificial intelligence. $3 194058
650 4 $a Information technology. $3 184390
653 $a Knowledge based AI
653 $a Knowledge extraction techniques
653 $a Natural language understanding
690 $a 0984
690 $a 0800
690 $a 0489
710 2 $a University of Rochester. $b Hajim School of Engineering and Applied Sciences. $3 708626
773 0 $t Dissertations Abstracts International $g 82-05B.
790 $a 0188
791 $a Ph.D.
792 $a 2020
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28092953