Learning to Tag from Open Vocabulary Labels
Identifieur interne : 000D85 ( Main/Exploration ); précédent : 000D84; suivant : 000D86Learning to Tag from Open Vocabulary Labels
Auteurs : Edith Law [États-Unis] ; Burr Settles [États-Unis] ; Tom Mitchell [États-Unis]Source :
- Lecture Notes in Computer Science [ 0302-9743 ] ; 2010.
Abstract
Abstract: Most approaches to classifying media content assume a fixed, closed vocabulary of labels. In contrast, we advocate machine learning approaches which take advantage of the millions of free-form tags obtainable via online crowd-sourcing platforms and social tagging websites. The use of such open vocabularies presents learning challenges due to typographical errors, synonymy, and a potentially unbounded set of tag labels. In this work, we present a new approach that organizes these noisy tags into well-behaved semantic classes using topic modeling, and learn to predict tags accurately using a mixture of topic classes. This method can utilize an arbitrary open vocabulary of tags, reduces training time by 94% compared to learning from these tags directly, and achieves comparable performance for classification and superior performance for retrieval. We also demonstrate that on open vocabulary tasks, human evaluations are essential for measuring the true performance of tag classifiers, which traditional evaluation methods will consistently underestimate. We focus on the domain of tagging music clips, and demonstrate our results using data collected with a human computation game called TagATune.
Url:
DOI: 10.1007/978-3-642-15883-4_14
Affiliations:
Links toward previous steps (curation, corpus...)
- to stream Istex, to step Corpus: 000765
- to stream Istex, to step Curation: 000765
- to stream Istex, to step Checkpoint: 000157
- to stream Main, to step Merge: 000D95
- to stream Main, to step Curation: 000D85
Le document en format XML
<record><TEI wicri:istexFullTextTei="biblStruct:series"><teiHeader><fileDesc><titleStmt><title xml:lang="en">Learning to Tag from Open Vocabulary Labels</title>
<author><name sortKey="Law, Edith" sort="Law, Edith" uniqKey="Law E" first="Edith" last="Law">Edith Law</name>
</author>
<author><name sortKey="Settles, Burr" sort="Settles, Burr" uniqKey="Settles B" first="Burr" last="Settles">Burr Settles</name>
</author>
<author><name sortKey="Mitchell, Tom" sort="Mitchell, Tom" uniqKey="Mitchell T" first="Tom" last="Mitchell">Tom Mitchell</name>
</author>
</titleStmt>
<publicationStmt><idno type="wicri:source">ISTEX</idno>
<idno type="RBID">ISTEX:0A645893CDD5762C7749FF9B0CDC78E44DBDB935</idno>
<date when="2010" year="2010">2010</date>
<idno type="doi">10.1007/978-3-642-15883-4_14</idno>
<idno type="url">https://api.istex.fr/document/0A645893CDD5762C7749FF9B0CDC78E44DBDB935/fulltext/pdf</idno>
<idno type="wicri:Area/Istex/Corpus">000765</idno>
<idno type="wicri:Area/Istex/Curation">000765</idno>
<idno type="wicri:Area/Istex/Checkpoint">000157</idno>
<idno type="wicri:doubleKey">0302-9743:2010:Law E:learning:to:tag</idno>
<idno type="wicri:Area/Main/Merge">000D95</idno>
<idno type="wicri:Area/Main/Curation">000D85</idno>
<idno type="wicri:Area/Main/Exploration">000D85</idno>
</publicationStmt>
<sourceDesc><biblStruct><analytic><title level="a" type="main" xml:lang="en">Learning to Tag from Open Vocabulary Labels</title>
<author><name sortKey="Law, Edith" sort="Law, Edith" uniqKey="Law E" first="Edith" last="Law">Edith Law</name>
<affiliation wicri:level="4"><country>États-Unis</country>
<placeName><settlement type="city">Pittsburgh</settlement>
<region type="state">Pennsylvanie</region>
</placeName>
<orgName type="university">Université Carnegie-Mellon</orgName>
</affiliation>
<affiliation wicri:level="1"><country wicri:rule="url">États-Unis</country>
</affiliation>
</author>
<author><name sortKey="Settles, Burr" sort="Settles, Burr" uniqKey="Settles B" first="Burr" last="Settles">Burr Settles</name>
<affiliation wicri:level="4"><country>États-Unis</country>
<placeName><settlement type="city">Pittsburgh</settlement>
<region type="state">Pennsylvanie</region>
</placeName>
<orgName type="university">Université Carnegie-Mellon</orgName>
</affiliation>
<affiliation wicri:level="1"><country wicri:rule="url">États-Unis</country>
</affiliation>
</author>
<author><name sortKey="Mitchell, Tom" sort="Mitchell, Tom" uniqKey="Mitchell T" first="Tom" last="Mitchell">Tom Mitchell</name>
<affiliation wicri:level="4"><country>États-Unis</country>
<placeName><settlement type="city">Pittsburgh</settlement>
<region type="state">Pennsylvanie</region>
</placeName>
<orgName type="university">Université Carnegie-Mellon</orgName>
</affiliation>
<affiliation wicri:level="1"><country wicri:rule="url">États-Unis</country>
</affiliation>
</author>
</analytic>
<monogr></monogr>
<series><title level="s">Lecture Notes in Computer Science</title>
<imprint><date>2010</date>
</imprint>
<idno type="ISSN">0302-9743</idno>
<idno type="eISSN">1611-3349</idno>
<idno type="ISSN">0302-9743</idno>
</series>
<idno type="istex">0A645893CDD5762C7749FF9B0CDC78E44DBDB935</idno>
<idno type="DOI">10.1007/978-3-642-15883-4_14</idno>
<idno type="ChapterID">Chap14</idno>
<idno type="ChapterID">14</idno>
</biblStruct>
</sourceDesc>
<seriesStmt><idno type="ISSN">0302-9743</idno>
</seriesStmt>
</fileDesc>
<profileDesc><textClass></textClass>
<langUsage><language ident="en">en</language>
</langUsage>
</profileDesc>
</teiHeader>
<front><div type="abstract" xml:lang="en">Abstract: Most approaches to classifying media content assume a fixed, closed vocabulary of labels. In contrast, we advocate machine learning approaches which take advantage of the millions of free-form tags obtainable via online crowd-sourcing platforms and social tagging websites. The use of such open vocabularies presents learning challenges due to typographical errors, synonymy, and a potentially unbounded set of tag labels. In this work, we present a new approach that organizes these noisy tags into well-behaved semantic classes using topic modeling, and learn to predict tags accurately using a mixture of topic classes. This method can utilize an arbitrary open vocabulary of tags, reduces training time by 94% compared to learning from these tags directly, and achieves comparable performance for classification and superior performance for retrieval. We also demonstrate that on open vocabulary tasks, human evaluations are essential for measuring the true performance of tag classifiers, which traditional evaluation methods will consistently underestimate. We focus on the domain of tagging music clips, and demonstrate our results using data collected with a human computation game called TagATune.</div>
</front>
</TEI>
<affiliations><list><country><li>États-Unis</li>
</country>
<region><li>Pennsylvanie</li>
</region>
<settlement><li>Pittsburgh</li>
</settlement>
<orgName><li>Université Carnegie-Mellon</li>
</orgName>
</list>
<tree><country name="États-Unis"><region name="Pennsylvanie"><name sortKey="Law, Edith" sort="Law, Edith" uniqKey="Law E" first="Edith" last="Law">Edith Law</name>
</region>
<name sortKey="Law, Edith" sort="Law, Edith" uniqKey="Law E" first="Edith" last="Law">Edith Law</name>
<name sortKey="Mitchell, Tom" sort="Mitchell, Tom" uniqKey="Mitchell T" first="Tom" last="Mitchell">Tom Mitchell</name>
<name sortKey="Mitchell, Tom" sort="Mitchell, Tom" uniqKey="Mitchell T" first="Tom" last="Mitchell">Tom Mitchell</name>
<name sortKey="Settles, Burr" sort="Settles, Burr" uniqKey="Settles B" first="Burr" last="Settles">Burr Settles</name>
<name sortKey="Settles, Burr" sort="Settles, Burr" uniqKey="Settles B" first="Burr" last="Settles">Burr Settles</name>
</country>
</tree>
</affiliations>
</record>
Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Musique/explor/OperaV1/Data/Main/Exploration
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 000D85 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Exploration/biblio.hfd -nk 000D85 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien |wiki= Wicri/Musique |area= OperaV1 |flux= Main |étape= Exploration |type= RBID |clé= ISTEX:0A645893CDD5762C7749FF9B0CDC78E44DBDB935 |texte= Learning to Tag from Open Vocabulary Labels }}
This area was generated with Dilib version V0.6.21. |