Ontology and the Lexiocn.week4

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
Lexical Resources Lexicalized Ontologies
. . . .
. . . . . .
Lab
Ontology and the Lexicon
Week 4: Ontological and Lexical Resources
Shu-Kai Hsieh
Lab of Ontologies, Language Processing and e-Humanities
GIL, National Taiwan University
March 18, 2014
Ontology and the Lexicon Shu-Kai Hsieh

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
..1 Lexical Resources
WordNet
FrameNet
..2 Lexicalized Ontologies
..3 Lab
Wiki Taxonomy
Chinese Wordnet

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Barsalou[1]
• We shall assume that concepts are people’s psychological
representations of categories (e.g., apple, chair); whereas
meanings are people’s understandings of words and other
linguistic expressions (e.g., ”apple”, ”large chair”).
• We shall argue that concepts and meanings diﬀer
substantially. Although they are related in important ways,
the relationship is one of complementarity, not equivalence.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
WordNet
WordNet
• A freely available lexical Database developed by Princeton
University.
• A semantic network which consists of synsets and relations.
• 155,287 word forms are grouped into 117,659 synsets
(WordNet 3.0)
• synsets are glossed, and interconnected with diﬀerent semantic
relations.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
WordNet
FrameNet
..3 Lab
Wiki Taxonomy
Chinese Wordnet

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
Frames: chunks of knowledge
..1 The notion of frame came up in the 1970s in AI and Cognitive
Science. Similar notions include schema, script and scenario.
..2 Words are deﬁned through the semantic roles they play and
the frames (types of event, relation, or entity) they evoke.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
source:
http://www.icsi.berkeley.edu/icsi/news/2014/02/charles-fillmore-dies-at-84

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
FrameNet
• Word evokes the frame.
• Instead of words, FN works with lexical units (LUs), each of
these being a pairing of a word with a sense.
Example (of FN work)
Let’s work through the Revenge frame following Fillmore (pp26-):
(https://framenet.icsi.berkeley.edu/fndrupal/sites/default/files/FNintroCJF.ppt); The
glossary also helps (https://framenet.icsi.berkeley.edu/fndrupal/glossary)

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
FrameNet Relations
The frames that we create, and thus the Frame Elements (FE) and
Lexical Units (LU) associated with them, are intended to be
situated in semantic space by means of frame-to-frame relations
and semantic types. The relations we use include Inheritance,
Subframe, Causative of, Inchoative of, and Using.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
FrameNet Relations
source: FrameNet (II) book
Inheritance An IS-A relation. The child frame is a subtype of the
parent frame, and each FE in the parent is bound to
a corresponding FE in the child. An example is the
Revenge frame which inherits from the Rewards and
punishments frame.
Subframe The child frame is a subevent of a complex event
represented by the parent, e.g. the Criminal
process frame has subframes of Arrest,
Arraignment, Trial, and Sentencing.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
FrameNet Relations
source: FrameNet (II) book
Using The child frame presupposes the parent frame as
background, e.g the Speed frame ”uses” (or
presupposes) the Motion frame; however, not all
parent FEs need to be bound to child FEs.
Perspective on The child frame provides a particular perspective
on an un-perspectivized parent frame. A pair of
examples consists of the Hiring and Get_a_job
frames, which perspectivize the Employment_start
frame from the Employer’s and the Employee’s point
of view, respectively.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
FrameNet Ontology

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
Lexical Resources: Comparison and Alignment
• FN: 800 frames, 10,000 LUs, exampliﬁed in more than
135,000 sentences.
• WN: 117659 synsets, 206941 word-sense pairs, ....

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
• WN: a syset comprises only synonyms of the same part of
speech; FN: a frame may include diﬀerent parts of speech,
and words with contradictory deﬁnitions (such as antonyms
related to the same idea).
• Statistical measure that shows where WordNet and FrameNet
agree well on the meanings of words and phrases, and where
they do not [2].

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
FrameNet
Out of the 67 annotated sentences that included the word curious,
48 both evoked the typicality frame (FN1) and used curious in the
WordNet sense of ”beyond or deviating from the usual or
expected” (WN1).

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Classiﬁcation of Ontologies
• Formal vs informal ontologies
• Lexicalized vs non-lexicalized ontologies

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
WN as a lexicalized ontology
• All of the WordNet noun synsets are organized into
hierarchies, with 25 top-level synsets headed by the unique
beginner synset entity.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
• E.g., Hypernymy/Hyponymy and Meronymy/Holonymy are
transitive and asymmetrical. So as Hyponymy generates a
hierarchical semantic structure, a hyponym inherits all the
features of the more generic concept and adds at least one
feature that distinguishes it from its superordinate. The
hierarchy is about 16 levels.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Multilingual Wordnets
• (EuroWordnet) The development of a multilingual database
with WordNets for several European languages, with 10,000
up to 50,000 synsets. (Dutch, German, French, Spanish,
Italian, Czech, Estonian).
• Inter-Lingual-Index, which are mainly based on EWN
synsets, serves as unstructured fund of concepts that provide
an efficient mapping across the languages;
• Various types of equivalence relations are distinguished to
link synsets with index records.
• Some cross-linguistic issues identified: different lexicalization,
differencs in synonymy and homonymy, etc.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Multilingual Wordnets
• (BalkaNet) (Romanian, Bulgarian, Turkish, Slovenian, Greek,
Serbian), with 10,000 synsets.
• (AsianWordnet) (Hindi, Indonesian, Lao, Mongolian,
Japanese, Burmese, Nepali, Sinhala, Thai, Vietnamese) 1
• (The Global WordNet Association, more than 50 languages) 2
1
http://www.asianwordnet.org
2
http://www.globalwordnet.org

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Figure : Multilingual WordNet Architecture: EWN’s model

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
The ideal Global Wordnet Model (Vossen)?
• Construct separate wordnets for each language.
• Contributors from each language encode the same core set of
concepts plus culture/language-speciﬁc ones.
• Synsets (concepts) are mapped cross linguistically via an
ontology, instead of just the English Wordnet.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
NLTK rocks

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Wiki Taxonomy
WordNet
FrameNet
..3 Lab
Wiki Taxonomy
Chinese Wordnet

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Wiki Taxonomy
Wiki: its role
• A good review of current state-of-arts can be found in [3].
• Resolving the Knowledge acquisition bottleneck: The
creation of very large knowledge bases has been made possible
by the availability of collaboratively-curated online resources
such as Wikipedia and Wiktionary.
• structured, semi-structured, unstructured resources.
• what are the advantages and disadvantages, respectively?

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Wiki Taxonomy
Wiki as semi-structured content for Ontologies
• Transforming Wikipedia into machine-readable knowledge
• Acquiring related terms: thesaurus extraction
• Relation extraction
• Leitmotif: generating semantics by exploiting the shallow
structure found in Wikipedia.
• Building and enriching ontologies from Wikipedia: YAGO,
WikiNet and BabelNet.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Wiki Taxonomy
Multilingual case of collaboratively-generated
semi-structured resources

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Wiki Taxonomy
• WikiTaxonomy (Ponzetto and Strube, 2007; Ponzetto and
Strube, 2011)(100k is-a relations)
• WikiNet: (Nas- tase et al., 2010; Nastase and Strube, 2013)
is a project which heuristically exploits diﬀerent aspects of
Wikipedia to obtain a multilingual concept network by deriving
not only is-a relations, but also other types of relations.
• MENTA (de Melo and Weikum, 2010), creates one of the
largest multilingual lexical knowledge bases by interconnecting
more than 13M articles in 271 languages.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Chinese Wordnet
Chinese Wordnet(s)
Existing CWNs:
• Sinica BOW (Huang et al, 2004)
• CWN.1 (Academia Sinica, 2004-2010) »> CWN.2
(NTU-Taiwan 2011-)
• COW (NTU-Singapore)
• Southeast University WordNet (SEW), 2008

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Chinese Wordnet
Chinese Wordnet(s)
Evaluation matters!
Example
• quantity w/o quality
• linguistic depth (co-predication, variants, etc)
• coarse or ﬁne-grained evalution methodology (e.g., via
annotators agreement or OntoClean methodology) (Hsieh,
2013)

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Chinese Wordnet
Chinese Wordnet.2
Launched and develped by AS (Chu-Ren Huang) and maintained
by LOPE lab (Shu-Kai Hsieh) at NTU.
• Successor of bilingual translation wordnet (Sinica BOW)
developed by Academia Sinica.3.
• condense in synsets, sparse in LSR.
• with some speciﬁc designs (e.g., meaning facets, paranymy,
etc).
• other variants 4
3
bow.sinica.edu.tw
4
googleCWN: lope.linguistics.ntu.edu.tw/gcwn

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Chinese Wordnet
Chinese Wordnet 2

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Chinese Wordnet
A wikiﬁed collaboration platform

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Chinese Wordnet
Homework
• 找一個角度，比較 WN 與 FN 的異同
• (修過 python 的同學，請多利用 nltk 來數量上的說明與論
據)

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Chinese Wordnet
Lawrence W Barsalou, Wenchi Yeh, Barbara J Luka, Karen L
Olseth, Kelly S Mix, and Ling-Ling Wu.
Concepts and meaning.
1993.
Gerard De Melo, Collin F Baker, Nancy Ide, Rebecca J
Passonneau, and Christiane Fellbaum.
Empirical comparisons of masc word sense annotations.
In LREC, pages 3036–3043, 2012.
Eduard Hovy, Roberto Navigli, and Simone Paolo Ponzetto.
Collaboratively built semi-structured content and artiﬁcial
intelligence: The story so far.
Artiﬁcial Intelligence, 194:2–27, 2013.
Sebastian Löbner.

.....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
....
.
....
.
.....
.
....
.
.....
.
....
.
....
.
.
. . . . . . . . . . .
. . . .
. . . . . .
Lab
Chinese Wordnet
Understanding semantics.
Routledge, 2013.

Ontology and the Lexiocn.week4

Empfohlen

Empfohlen

Weitere ähnliche Inhalte

Ähnlich wie Ontology and the Lexiocn.week4

Ähnlich wie Ontology and the Lexiocn.week4 (20)

Kürzlich hochgeladen

Kürzlich hochgeladen (20)

Ontology and the Lexiocn.week4