not-so-linked solution to the linked data mining challenge 2016
TRANSCRIPT
![Page 1: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/1.jpg)
Not-So-Linked Solution to theLinked Data Mining Challenge 2016
Jedrzej Potoniec
Institute of Computing Science, Poznan University of Technology
30.05.2016
![Page 2: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/2.jpg)
Agenda
1 Feature construction
2 Machine learning workflow
3 Insight into ML model
4 Short conclusion
![Page 3: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/3.jpg)
Linked datasets: DBTune
![Page 4: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/4.jpg)
Linked datasets: DBTune
![Page 5: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/5.jpg)
Linked datasets: DBpedia
https://en.wikipedia.org/wiki/Strange_Mercy?oldid=667760297
![Page 6: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/6.jpg)
Linked datasets: DBpedia
https://en.wikipedia.org/wiki/Strange_Mercy?oldid=667760297
![Page 7: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/7.jpg)
Linked datasets: DBpedia
https://en.wikipedia.org/wiki/Strange_Mercy?oldid=667760297
![Page 8: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/8.jpg)
Linked datasets: DBpedia
https://en.wikipedia.org/wiki/Strange_Mercy?oldid=667760297
![Page 9: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/9.jpg)
Linked datasets: DBpedia
http://dbpedia.org/resource/Strange_Mercy
![Page 10: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/10.jpg)
Linked datasets: DBpedia
http://dbpedia.org/resource/Strange_Mercy
![Page 11: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/11.jpg)
Linked datasets: DBpedia
http://dbpedia.org/resource/Strange_Mercy
![Page 12: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/12.jpg)
Non-linked datasets
![Page 13: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/13.jpg)
Wikipedia
{{album ratings
|MC=85/100<ref name=metacritic/>
| rev1 = [[AllMusic]]
| rev1Score = {{rating|4|5}}<ref name=allmusic>Phares, Heather. [http://allmusic.com/album/strange-mercy-r2220859/review Strange Mercy St. Vincent]. [[Allmusic]]. Retrieved 14 September 2011.</ref>
| rev2 = ’’[[The A.V. Club]]’’
| rev2Score = A<ref name=avclub>Adams, Erik. [http://www.avclub.com/articles/st-vincent-strange-mercy,61595/ St Vincent: Strange Mercy]. [[The A.V. Club]]. 13 September 2011. Retrieved 13 September 2011.</ref>
| rev3 = [[Consequence of Sound]]
| rev3Score = {{rating|4.5|5}}<ref name=cos>Kivel, Adam. [http://consequenceofsound.net/2011/09/album-review-st-vincent-strange-mercy/ Album Review: St. Vincent Strange Mercy]. [[Consequence of Sound]]. 9 September 2011. Retrieved 9 September 2011.</ref>
| rev4 = ’’[[The Guardian]]’’
| rev4Score = {{rating|4|5}}<ref name=guardian>Nicholson, Rebecca. [http://www.guardian.co.uk/music/2011/sep/08/st-vincent-strange-mercy-review St Vincent: Strange Mercy review]. [[The Guardian]]. 8 September 2011. Retrieved 8 September 2011.</ref>
| rev5 = ’’[[The Observer]]’’
| rev5Score = {{rating|5|5}}<ref name=observer>Woodcraft, Molloy. [http://www.guardian.co.uk/music/2011/sep/11/st-vincent-strange-mercy-review St Vincent: Strange Mercy review]. [[The Observer]]. 11 September 2011. Retrieved 12 September 2011.</ref>
| rev6 = [[Pitchfork Media|Pitchfork]]
| rev6Score = 9.0/10<ref name=pitchfork/>
| rev7 = [[PopMatters]]
| rev7Score = 9/10<ref name=popmatters>Pan, Arnold. [http://www.popmatters.com/pm/review/148401-st.-vincent-strange-mercy/ St. Vincent: Strange Mercy]. [[Popmatters]]. 12 September 2011. Retrieved September 2011.</ref>
| rev8 = ’’[[Q (magazine)|Q]]’’
| rev8Score = {{rating|4|5}}<ref name="q"/>
| rev9 = [[Slant Magazine]]
| rev9Score = {{rating|4|5}}<ref name=slant>Liedel, Kevin. [http://www.slantmagazine.com/music/review/st-vincent-strange-mercy/2616 St. Vincent: Strange Mercy]. [[Slant Magazine]]. 12 September 2011. Retrieved 12 September 2011.</ref>
| rev10 = ’’[[Spin (magazine)|Spin]]’’
| rev10Score = 9/10<ref name=spin>Anderson, Stacey. [http://www.spin.com/reviews/st-vincent-strange-mercy-4ad St. Vincent ’Strange Mercy’]. [[Spin (magazine)|Spin]]. Retrieved 7 September 2011.</ref>
}}
![Page 14: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/14.jpg)
Musicbrainz
![Page 15: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/15.jpg)
Discogs
![Page 16: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/16.jpg)
Amazon
![Page 17: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/17.jpg)
Non-linked dataset
RDF github.com/jpotoniec/LDMC2016/blob/master/data.ttl
CSV
github.com/jpotoniec/LDMC2016/blob/master/wikiscraper/
reviews.csv
github.com/jpotoniec/LDMC2016/blob/master/musicbrainz/
ratings.csv
github.com/jpotoniec/LDMC2016/blob/master/discogs/
discogs.csv
github.com/jpotoniec/LDMC2016/blob/master/amazon/amazon/
result.csv
![Page 18: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/18.jpg)
Machine learning workflow
normalization (Z-transformation)
missing values replacement
logistic regression
cross-validation
Wikipedia+MusicBrainz+Discogs+Amazon=91.7 ± 2.17%
github.com/jpotoniec/LDMC2016/blob/master/RapidMiner/learn.rmp
![Page 19: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/19.jpg)
Machine learning workflow
normalization (Z-transformation)
missing values replacement
logistic regression
cross-validation
Wikipedia+MusicBrainz+Discogs+Amazon=91.7 ± 2.17%
github.com/jpotoniec/LDMC2016/blob/master/RapidMiner/learn.rmp
![Page 20: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/20.jpg)
Machine learning workflow
normalization (Z-transformation)
missing values replacement
logistic regression
cross-validation
Wikipedia+MusicBrainz+Discogs+Amazon=91.7 ± 2.17%
github.com/jpotoniec/LDMC2016/blob/master/RapidMiner/learn.rmp
![Page 21: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/21.jpg)
Machine learning workflow
normalization (Z-transformation)
missing values replacement
logistic regression
cross-validation
Wikipedia+MusicBrainz+Discogs+Amazon=91.7 ± 2.17%
github.com/jpotoniec/LDMC2016/blob/master/RapidMiner/learn.rmp
![Page 22: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/22.jpg)
Machine learning workflow
normalization (Z-transformation)
missing values replacement
logistic regression
cross-validation
Wikipedia+MusicBrainz+Discogs+Amazon=91.7 ± 2.17%
github.com/jpotoniec/LDMC2016/blob/master/RapidMiner/learn.rmp
![Page 23: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/23.jpg)
Machine learning workflow
normalization (Z-transformation)
missing values replacement
logistic regression
cross-validation
Wikipedia+MusicBrainz+Discogs+Amazon=91.7 ± 2.17%
github.com/jpotoniec/LDMC2016/blob/master/RapidMiner/learn.rmp
![Page 24: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/24.jpg)
Machine learning workflow
normalization (Z-transformation)
missing values replacement
logistic regression
cross-validation
Wikipedia+MusicBrainz+Discogs+Amazon=91.7 ± 2.17%
github.com/jpotoniec/LDMC2016/blob/master/RapidMiner/learn.rmp
![Page 25: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/25.jpg)
Attribute weights
attribute coefficient
review score from Pitchfork 2.859review score from AllMusic 2.437review score from Stylus 1.926number of people owning an album according to Discogs 1.465review score from Entertainment Weekly 1.274review score from The Guardian 1.096. . .number of reviews on Amazon −0.442
github.com/jpotoniec/LDMC2016/blob/master/RapidMiner/model.ioo
![Page 26: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/26.jpg)
Conclusions
The Semantic Web: are we there yet?
![Page 27: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/27.jpg)
What-if
github.com/jpotoniec/LDMC2016/tree/master/LOD/
DBpedia
accuracy 76.02% ± 1.98%
highest weight 0.392 fordbp:label/dcterms:subject=dbc:Indie rock record labels
DBpedia+non-linked dataset
accuracy 86.74% ± 2.09%
learning time 25m27s
highest weight 1.177 for review score from Pitchfork
![Page 28: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/28.jpg)
What-if
github.com/jpotoniec/LDMC2016/tree/master/LOD/
DBpedia
accuracy 76.02% ± 1.98%
highest weight 0.392 fordbp:label/dcterms:subject=dbc:Indie rock record labels
DBpedia+non-linked dataset
accuracy 86.74% ± 2.09%
learning time 25m27s
highest weight 1.177 for review score from Pitchfork
![Page 29: Not-So-Linked Solution to the Linked Data Mining Challenge 2016](https://reader031.vdocument.in/reader031/viewer/2022030318/58f35a201a28ab396b8b458f/html5/thumbnails/29.jpg)
What-if
github.com/jpotoniec/LDMC2016/tree/master/LOD/
DBpedia
accuracy 76.02% ± 1.98%
highest weight 0.392 fordbp:label/dcterms:subject=dbc:Indie rock record labels
DBpedia+non-linked dataset
accuracy 86.74% ± 2.09%
learning time 25m27s
highest weight 1.177 for review score from Pitchfork