Friday, December 25, 2015

Simulating Joyce's prose

joycean word-processor
helps you write more like joyce

episode styles
exercise: rewrite episode A in style B

obscure word lists

patterns of interior monologue vs dialogue&c
classes of internal thought (memory, sensation, fantasy)

generate path markers
pick a character, generate a path, generate a monolog
intersect multiple simultaneous paths (ch10)

motifs: figures of speech
eg collect 'fire' sayings and outline a fire-themed episode

speech patterns
extrapolate similar phrases based on sample from Ulysses

train a neural net on each set of authors in oxen [cf]

Matt Schneider suggests ELIZA as a model for Ithaca's q&a, Maybe a (no-graphics-style) text adventure that delivers canned text if you guess the right question?

a simple algorithm could use color to display how cliched each word-combination is, as you read or write.

Wednesday, December 23, 2015

Text animation

can we enhance ulysses by animating the text?

cf breaking up penelope into short lines??

Matt Schneider suggests ELIZA as a model for Ithaca's q&a,maybe with the question appearing a letter at a time, and the answer flowing back after a pause?

Sunday, December 20, 2015

Generating monologs

more than any previous book, ulysses is a compilation of monologs along paths.

it's easy to generate random pseudo-joycean monologs

refining them is an infinite challenge

 one possibility is to time the output to approximate realtime

one approach is to classify the kinds of thoughts each character has at each moment,
and generalise those patterns

create a simplified ulysses where characters' monologs consist entirely of a small fixed set of cliches of each type

compile list of things-to-see with gps positions

choose path and compile sequence of things along path

different characters may have different memories/associations to things
good memories/ bad memories

internal states like hunger, thirst, anxiety, lust,
randomly noticed, or systematically

daily to-do list for each character
tendency to dillydally
monitoring time

financial calculations
budget for self-indulgence

encountering people
names, ages, families and occupations from census
allowing oneself time to socialise
ask about housemates


Tuesday, December 15, 2015

Thinking about sex

Each character has a level of libido that varies according to obscure rules: cycles, times of day, stimulation, boredom, flattery...

Each person they meet or think of has some level of attractiveness, accessibility, risk...

(Prostitutes may be attractive and accessible but risky wrt stds. Clergymen may be attractive but inaccessible.)

We see Stephen choosing a semi-random image to fantasize about.

Bloom understands Martha is just an amusing game he'll never really dare to meet.

Drinking lowers standards, increases accessibility.

Thursday, December 10, 2015

WikiData knowledgebase invites anyone to flesh out their mirror of Wikipedia's knowledgebase, eg by adding additional Ulysses characters, and spelling out the links between characters.

Their representation language is still quite limited, and discourages shortcuts like representing a character's spouse as a simple name-string instead of instantiating an entire new fictional-human item.

There's lots of colorful details Joyce introduces to define character, but few of these fit neatly into existing categories: occupation, 'uses', nickname, member of, educated at, significant person...

Ideally we'll be able to extract useful 'views' of the data, down the line.

One tricky thing is differentiating real people from their fictional doubles: is 'Cranly' identical to JF Byrne?

Is Shakespeare 'present in' chapter 9?