I want to:
- the weight of rain
- expected relationships, rigorously shown
- meaningful adjacencies
- data in a way that promotes intuition and immediate emotional response
- explain the domain area and some cultural factors/situations/unique aspects about it
- artistic delight
- highlight unexpected connections & context with the rest of the world
- it's ok to be subjective/focused with the visualization!
I do not want to:
- just count things
- just encode things in attractive but opaque iconography
- have too much data to the point that occludes meaning
Feedback from Golan:
- don't be afraid to have a point of view: make the choice to look at a single topic or theme
- brand names? consumerism in music, cultural zeitgeist
- emotional analysis
- love and heartbreak
- just investigate one poet/artist and their preoccupations with:
- love
- revenge
- branding
- emotions and emotional spread
- men vs women: how much do men talk about women? men about men? women about women/men? Bechdel test
- be specific, like the body parts of lyrics (also note how the rebus things show just body parts and other words as dots)
- generating the text is a lot of work and if the purpose is to display differences between things, this doesn't cut it
- narrow down the dta set to find something specific and interesting
- mash up poetry and rap (found poetry, use quotes or k-grams)
- visualize graph of generated text?
- how to connect poems given limited data?
- small multiples to compare corpora
- ratio of nouns to verbs
- ratio of nouns and verbs to other things
- positive versus negative words
- swearing
- sexual content
- pronouns: I/you/me
- average content length (words/lines)
- average line length (syllables/words)
- villanelles out of song lyrics from different genres, artists
- sentiment analysis and literacy analysis
- find most common nouns/verbs as endings
- use TV Tropes information (here’s Kanye’s page)
- blackout poetry
- visual presentation predicated by metadata/beat/&c of the music
- remix poetry and music? poetry and rap?
- find most similar artists
- Rap Genius
- Poetry Genius
- select particular artists and poets?
- scrape particular decades
- Scrape from Poetry Foundation
- Scrape from Poemhunter
- NLTK
- Wordnet through NLTK—use the synonyms?
- CMU Dict: gives lexical stresses and syllables?
- rhyming dictionary: scrape/query Rhymezone? also has synonyms but are not super useful
- Regressive Imagery Dictionary
- Linguistic Inquiry & Word Count
- lyric lines too short? connect them with
- how to deal with sparse n-grams?
- Yusef Komunyakaa
- Pablo Neruda
- Regina Spektor
- Arcade Fire
- Theophilus London
- Aesop Rock
- Kanye West
- Azealia Banks
- Notorious B.I.G.
- Scrape for sets of poems and lyrics (just title/author/text)
- for lyrics, ignore [Hook], [Verse 2], x2, (x2) and similar
- Process to a good format (individual text files for each)
- display where each phrase comes from (song + artist/poem + poet)
- mood of songs
- morph poetry styles
- which poetic forms?
- tweet couplets: one from poem, one from song; one from rap
- clear juxtaposition/genre identity
- make a Twitter account out of it
- rap haikus
- rhyming connects unexpected sources together
- @pentametron
- blackout poetry and similarly, humament poetry—
- isolate emotions?
- i
- verbatim found poetry
- new poetry forms