Visualization of Word Meanings Using Formal Concept Analysis

Visualization of Word Meanings Using Formal Concept Analysis and Concept Lattices L John Old School of Computing Napier University Edinburgh, Scotland 1

• Roget’s Thesaurus – Structure of Roget’s Thesaurus – Using FCA to visualize and explore semantic relationships within Roget’s Thesaurus • Indo-European roots – Using FCS to explore structural relationships between roots – Using FCA to explore relationships between the etymological descendants of roots within Roget’s Thesaurus 2

Roget’s Thesaurus Conceptual Hierarchy Class Roman sub-class (R-Class) Letter sub-class (L-Class) Category Paragraph: [POS] Sense: {Entry, entry, . . . [Crossref]} Sense Index (Word Index) 3

Roget Top Level Classes (1963 Edition--RIT) 1. Abstract Relations 2. Space (dimensions) 3. Physics (physical properties) 4. Matter (materials) 5. Sensation (senses) 6. Intellect 7. Volition (motivation) 8. Affections (emotions) 4

Under the Hierarchy: words are polysemous 5

Conceptual Hierarchy - Senses of “over” (11 of 22) Abstract Relations Relation Quantity Absolute Comparitive Conjunctive Whole. Relation Quantity someness Order Number Consecutive Indeterminate Order Number Addition Relation Superiority 40 Remainder Completeness 43 9 36 56 Time Absolute Time End Repetition Time 70 103 105 Power Relative Time Power in General Posteriority Impotence 117 157 The Past 119 6

Sense Index CATEGORY PARAGRAPH 40. Addition NOUNS 1. … Index word SENSE ADVS. 10. and also, additionally, in addition, and all [coll. ], yet, extra, on the side [slang], moreover, farther, further, furthermore, overplus, again, more, similarly, likewise, by the same sign, by the same token, item, also, then, to boot, and so, as well, too, else, au reste [F. ], besides, plus, into the bargain, past, on top of, over, above, over and above, at that [coll. ], beyond, beside; . . . PREPS. etc. 11 with, including. . . ENTRY=40: 1 -over 7

Word Index WORD over Index word verbs … advs. additionally 40: 10 throughout 56. 18. . . preps. concerning 9. 12 exceeding 36. 20 . . . on 227. 40 . . . PARAGRAPH CATEGORY OVER AGAINST … overabound. . . ENTRY=40: 1 -over 8

Word-Sense Relation Synonymy SENSES WORDS Polysemy ENTRY=40: 1 -over 9

Senses and Synonyms of SEA (synset level) - 34. 3. 4 - 394. 14. 2 - 396. 1. 1 sea 34. 3. 4 sea, oceans, worlds… 394. 14. 2 sea, heavy swell. 396. 1. 1 sea, ocean, mere … 10

mere Senses of the Synonyms of SEA (synset level) 11

“mere” nested line diagram (full hierarchy) 12

Senses (8) and synonyms (49)of ”agitate” 13

GIS “Lattice map” 14

Web Map Interface 15

Map Server Query 16

• Using FCA to visualize Indo. European root relationships – Inter-root structural relationships – Root derivatives (descendants) in Roget’s Thesaurus 17

Grimm’s Law GHESLO- => kilo SKRIBH- BHEL-3 => scribe, enscribe, script, … => bloom, … flower DHE- DHES- => do, … thesis => deus, … theology 18

Consonant Elements SKRIBH- {cut, separate} (S)GH-R(-)-BH GHREBH KREUP- {scab} GH-R(--)-BH GHREBH • Elements • • • GH BH DH R, L M, N W • (S) • -(AEIOU)- 19

IE Root Meaning BHERDHGHEL-3 GHER-4 GLEUBHKES-1 Kel-1 MAI-1 PEIG-1 PEUREI-1 SEKSKEISKEL-1 (S)KEPSKER-1 SKREUSKRIBHSMISWER-3 TEM- cut scratch, carve, cut, cleave cut strike, cut cut, mark by incision cut, stamp, strike cut, scratch, tear cut, split, separate cut, scrape, hack cut, cutting tool cut, separate cut, work with a sharp tool cut, pierce cut Indo European Roots meaning “cut” (S)GHE 20

Consonant Element Classification of Roots meaning “cut” 21

“Prototypical” words descended from “cut” roots Root English #Descendants BHERDHGHEL-3 GHER-4 GLEUBHKEL-1 KES-1 MAI-1 PEIG-1 PEUREI-1 SEKSKEISKEL-1 (S)KEPSKER-1 SKREUSKRIBHSMISWER-3 TEM- BOARD GELD GASH CLEAVE GLADIATOR CASTRATE MAIM PIGMENT AMPUTATE REAP SICKLE SHED SCULPT SHAPE SCAR SHRED SCRIBE SMITH SWORD ATOM 3 1 2 3 9 4 5 6 9 7 11 12 12 11 24 5 10 1 1 8 22

English words derived from IE Roots meaning “cut” 23

Sek- a close-up of one “cut” root • SEK- > extended roots skei-, sker-1 (Pokorny 2. sk- 895) • Descendant Words: {dissect, intersect, notch, saw, scythe, section, sector, segment, sickle, skin} • 11 words in 54 senses in 36 RT Categories 24

All Categories of senses of descendants of IE root SEK- 25

Restricted Categories of senses of descendants of SEK- (Categories with at least 2 SEK- descendants) 26

END 27

Cognates – De/Eng DE Brett clever Ameise Feile amputieren reif Insekt Sektor Sense Sichel Schi Schale scharf Schrot Schmied Anatomie ROOT BHERDHGLEUBHMAI-1 PEIG-1 PEUREI-1 SEKSEKSKEISKEL-1 SKER-1 SKREUSMITEM- ENG BOARD CLEAVE ANT FILE AMPUTATE RIPE INSECTOR SCYTHE SICKLE SKI SHELL SHARP SHRED/scrap SMITH ANATOMY 28

Compounding • Wheel (circle, …) < kw(e)-kwl-o- < kwel + kwel < KWEL-1 [ = turn, move around, revolve] • Quarter (four, …) < OE fower < Gc *fe(d)wor- < KWETWER- < KWET- + WER[ = poss KWET- strike, shake; WER-3 turn, bend ] 29