vango project
TRANSCRIPT
Van GoYour personal art curator
Zuzanna Klyszejko
Problem When you’re visiting a new place your time is limited
There is too much to see
You need to make a choice
If a visitor wants to see only landscapes and plants, there should be a way to choose a subset of museum
objects that makes the most of their time
”This painting was said by him to have been inspired by
the work of Li Cheng, a tenth-century landscape artist.
Its spare, rather dry brushwork again repeats the
deliberately simple, austere quality that is the feature
of many so-called literati paintings. Hanging scroll.
Landscape. Bare trees in winter, with reference to Li
Cheng (919-67) and a river. Painted in a very dry style.
Inscriptions and seals. Ink on paper.”
Scraped 100 000 curatorial descriptions of paintings and drawings from British Museum’s database
”this painting be say by him to have be inspire by the work of li cheng , a tenth - century landscape artist . its spare , rather dry brushwork again repeat the deliberately simple , austere quality that be the feature of many so - call literati painting . hang scroll . landscape . bare tree in winter , with reference to li cheng ( 919 - 67 ) and a river . paint in a very dry style . inscription and seal . ink on paper.”
Tokenizing and Lemmatizing
Scraped 100 000 curatorial descriptions of paintings and drawings from British Museum’s database
Tokenizing and Lemmatizing
Count Vectorizer + TFIDF
Scraped 100 000 curatorial descriptions of paintings and drawings from British Museum’s database
Tokenizing and Lemmatizing
Count Vectorizer + TFIDF
Latent Semantic Analysis (PCA)
Scraped 100 000 curatorial descriptions of paintings and drawings from British Museum’s database
Tokenizing and Lemmatizing
Count Vectorizer + TFIDF
Latent Semantic Analysis (PCA)
Cosine similarity
Scraped 100 000 curatorial descriptions of paintings and drawings from British Museum’s database
How does it work in practice? Demo
Problem: Validation
Problem: Validation
Landscapehut
lake
cliff
mountainamid
bridge
cattledistance
Landscape
hut
tree
lake river
cliff
mountainamid
bridge cattle
bank
distance
stream
Landscape
boat
fish
wind
windmill
Landscape
hut
lake river
cliff
mountainamid
bridge cattle
distance
Landscape
hut
tree
stream
lake river
cliff
mountainamid
bridge cattle
bank
distance
Landscape
boat
fish
wind
windmill
hut
tree
stream
lake river
cliff
mountainamid
bridge cattle
bank
distance
boat
fish
wind
windmill
Landscape Landscape
treeriver
cliff
mountain
cattle
bank
JACCARD INDEX
Dataset 1
Train a model
Dataset 2
Train a model
Dataset 3
Train a model
K-fold cross validation to verify the model using Jaccard index
“landscape”
About me
PhD in Cognitive Neuroscience (NYU)
VanGo web app: vango.hopto.org
Zuzanna Klyszejko