Continuing Adventures in Topic Modeling Infinite Jest

Sat, Aug 11, 2012

I thought I’d continue my experiments in topic modeling with Infinite Jest. A twenty-topic Gibbs sampling turned up such free-associative gems as: {“alarm,” “fur,” “nude,” “matrix,” “rehab,” “warns,” “hallway,” “underwear,” “hispanic,” “beach”}, which I’d argue is a pretty good thematic summary. (I especially love that “matrix” showed up there; bless the randomized oracular algorithm and its serendipitous discoveries.)

Anyway, the full list is below. Sorry for the formatting.

Topic 1 Topic 2 [1,] "alarm" "granted" [2,] "fur" "blocked" [3,] "nude" "anticipate" [4,] "matrix" "paint" [5,] "rehab" "guilty" [6,] "warns" "nurture" [7,] "hallway" "lord's" [8,] "underwear" "lurid" [9,] "hispanic" "piss" [10,] "beach" "check" Topic 3 Topic 4 [1,] "leisure" "context" [2,] "booth" "valuables" [3,] "hiked" "distress" [4,] "descend" "petropolis" [5,] "darts" "mark" [6,] "fate" "pure" [7,] "lottery" "bldg" [8,] "raison" "realizing" [9,] "ironed" "logic" [10,] "designer" "stern" Topic 5 Topic 6 [1,] "dropping" "smothergill" [2,] "required" "teaching" [3,] "utilize" "openness" [4,] "compromised" "hawaiian" [5,] "she'll" "abstraction" [6,] "feature" "millicent's" [7,] "pour" "bowls" [8,] "homosexual" "stations" [9,] "undistinguished" "hesitant" [10,] "invitational" "tend" Topic 7 Topic 8 [1,] "admission" "perform" [2,] "casually" "tones" [3,] "really" "january" [4,] "stink" "scalp" [5,] "thorax" "bush" [6,] "suspects" "drip" [7,] "decade" "high-resin" [8,] "soma" "discussed" [9,] "acquire" "ascend" [10,] "toothless" "km" Topic 9 Topic 10 [1,] "protect" "gathering" [2,] "helped" "vast" [3,] "straightened" "dismantled" [4,] "birthday" "fatal" [5,] "untitled" "panels" [6,] "kindness" "prostitute" [7,] "named" "intellectual" [8,] "sucking" "newspaper" [9,] "troubling" "bangs" [10,] "frustration" "trained" Topic 11 Topic 12 [1,] "unspeakable" "legend" [2,] "dependence" "jennie" [3,] "temples" "amused" [4,] "wept" "dart" [5,] "magic" "arrayed" [6,] "tapered" "deeper" [7,] "planes" "delivering" [8,] "knuckle" "supreme" [9,] "wiping" "gardens" [10,] "swell" "sooty" Topic 13 Topic 14 [1,] "wears" "subtly" [2,] "dial" "spectral" [3,] "theaters" "fuzzy" [4,] "blowing" "vividly" [5,] "roach" "steaming" [6,] "colleague" "caution" [7,] "latin" "fire" [8,] "psychology" "pronounce" [9,] "tear" "rhetorically" [10,] "citizens" "neglect" Topic 15 Topic 16 [1,] "overwhelming" "separatism" [2,] "philosophical" "strategy" [3,] "altered" "lowered" [4,] "go" "surprisingly" [5,] "formality" "m.s" [6,] "devout" "bodies" [7,] "howls" "shine" [8,] "crop" "materials" [9,] "aware" "yyy" [10,] "forcing" "apologies" Topic 17 Topic 18 [1,] "one" "alien" [2,] "gately" "promise" [3,] "it's" "entrepreneurial" [4,] "time" "spends" [5,] "hal" "anhedonia" [6,] "right" "packet" [7,] "here" "ellipse" [8,] "head" "fide" [9,] "kind" "vertical" [10,] "see" "parents" Topic 19 Topic 20 [1,] "regions" "work" [2,] "wonder" "forever" [3,] "observation" "anyplace" [4,] "vinyl" "radial" [5,] "performing" "whacked" [6,] "risk" "considering" [7,] "bowed" "addict" [8,] "washing" "effects" [9,] "jet" "concept" [10,] "seeking" "chu's"