ATR and Poetry: Li Bai & Du Fu

• The presentation discusses the application of terminology through a case study, focusing on unveiling the artistic unconsciousness of the Chinese poets Li Bai and Du Fu by term comparison.

Automatic Term Recognition (ATR) by Corpus Comparison

• Corpus is a collection of texts
• Observation: Terms are of greater prominence in a Language for Specific Purposes (LSP) than in a Language for General Purposes (LGP)
• Terms are more prominent in a subject field compared to the public/general domain.
• Terms are more prominent in specialized articles than in general ones.
• Terms are more noticeable in a text of special topic compared to a general topic.
• A word with a higher frequency in a subject field corpus compared to its frequency in the public domain is more likely to be a term in that subject field.
• Termhood = prominence = frequency difference

Comparative Studies of Li & Du's Poetry

• Li Bai and Du Fu are considered two of the greatest Chinese poets.
• They represent the two peaks of Tang poetry described as "Brightest Gemini / twin stars".
• Everyone knows them and their poem is great.
• Their poetry differs significantly in content and style.
• There are many existing studies on Li Bai and Du Fu's poetry.
• Most critics show a strongly biased preference of one over the other.

Previous Research on the Poets

• Yan Yu (Song dynasty) stated "Li's highness is beyond Du & Du's profoundness beyond Li".
• Li Bai is the greatest romanticist; pursuing “immortal being" and known as "poetic immortal".
• Du Fu is the greatest in realist; full of profoundness, toughness, mercy, worry, and sadness; best known as "poetic saint"

Problems with Existing Studies

• Countless texts exist comparing their poetry.
• Most studies are based on subjective reading experiences, with few examples.
• There is a need for a holistic view with empirical evidence due to current subjective preferences.
• Examples are used as isolated evidence, lacking comprehensive, systematic, and reliable qualities.
• Their poetry shares many common characteristics.
• There has been no substantial progress after Yan Yu.

Reading: Form to Meaning

• Many forms exist in language.
• Form in language can be defined as a sound unit or fragment.
• The smallest form of sound = Phone (Phoneme)
• The smallest written form = Morph
• The smallest meaningful form = Morpheme
• The minimum free form = Word
• Which words are terms?
• Structure includes string, tree, graph, etc.
• Relations link these structures.

Reading: Form to Meaning of Language

• Every language form carries certain meaning or function.
• Language forms include morpheme, word, phrase, sentence, paragraph, discourse/document, etc.
• The meaning of a poem is one concept.
• The meaning of Li Bai's poem collection as another matter.
• The meaning of the corpus.
• Representation of this meaning is needed.
• Formalism is crucial for studying language.
• A 'macro' form needs to be used.

Frequency Statistics as a Simple Macro Expression

• Each text (string) can be converted into a "word-bag".
• A word-bag model offers: an unordered set of words, with frequency rates.
• A vector characterizes with words.
• A simpler model exists within the character bag model: character frequency statistics, with the character vector =? Meaning?

Objectives of Comparison: Seeking Commonalities and Differences

• Exploring which aspects or parts are the same and similar determines how similar they are, and requires a measure of similarity.
• Exploring which aspects or parts are different and opposite reveals the extent of the differences, and necessitates a measure of dissimilarity.
• The parts that are the same are the basis of the comparison.
• The parts that are different are the result of the comparison.
• Different ≠ Opposite
• Antonyms have the most similar semantics.

Frequency Comparison of Li and Du's Collected Poems

• Seeking sameness, difference.
• "Not"/"no," related to negation, questions.
• "People," is the core of concern.
• References to "mountains, clouds, wind, sun, moon, water, rivers, seas" represent the natural environment and space for survival.
• Mentions of "day, month, year, time, spring, autumn, now, old" denote time, seasons, and years.
• Phrases related to "come, go, return, travel, guest" describe movement and travel.

Word frequency distribution in Li Bai and Du Fu

• Sentiment analysis
• Du Fu often wrote old, sick, rainy, ruined, desolate, ravine; chaotic horses, late, small villages, deep, bleak, and cold.
• Li Bai mentions moon, sky, sea, clouds, mountains, bright water, green; laughter, flowing, void, flying, clear songs, sunny, up, dragon, wind; wine, pines, purple, and calm.
• Personality determines fate.
• Du Fu died on a sick boat on the river, while Li Bai died by drowning trying to embrace the moon.
• What caused these differences?

Categorical Analysis

• Natural scenery as themes include rivers, rain, fields, gorges, horses, dusk, poetry heads, villages, books, boats, cities, woods, and late in Du Fu, versus flowers, tombs, pine winds, jade, water, mountains, clouds, gold seas, sky, and moon in Li Bai.
• Themes of people, landscapes and relationships include messy poetry, villages, books, officials, city states, revered childre, official mansion, soldiers, thieves, children, palace halls, county roads, scholars, fishermen, temples, and lords from Du Fu, versus jade palaces, embroidered buildings, windows, temples, strings, Yao Dan, drinking glasses, zithers, dancing girls, swords, buildings, terraces, celestial beings, wine, songs, people, and lords in Li Bai.
• Closed class words are "oneself / themselves, not yet, already, more, have to / need to, must / should, still (yet), always / ever, still (yet) towards, finally, you also/even, why / how, those are / of, I should / ought to, combine, correct, continuously, very/extremely, 100, not, moreover" found in Du Fu, as well as "who, yet / but, you, use by, 60,000, nine, two, under, then, return, above, three, as, together - in, five, can / be possible to, how can / why, 1,000, if, this/thus/such, not apply to, of me/mine, lord" in Li Bai.

Further Categorical Analysis

• Light, Color, Temperature in Du Fu included cold light, darkness, green, cold, darkness, hot, cool, ice; while in Li Bai, there was light, green, white, violet, green, and bright gold.
• Animals and plants included horses, crutches, reeds, bamboo roots, thorns, U-shaped bird's nests, Eucommia leaves, pavilions, Ju (chrysanthemum) Du, shoe-like, snakes from Du Fu; where as Li Bai had trees, moss, parasol trees, cinnamon, chickens, swans, penglai, Lu Wei, phoenix, cranes, tigers, monkeys, wormwood, peach blossoms, and fairy pines.
• On psychology, physiology, sensation, movement in Du Fu there was old age, sickness, weakness, heads, looks, cold, happiness, eyes, body, questions, worries, sorrows, sweetness, dread, pity, hearings, needs, love, faces, crying out blood, walking coldly, and speechlessness, from Du Fu versus sitting, viewing, sorrow, drinking, seeking, talk, appearing, shyness, waving, shaking, joy, sighing, hoping, offering, giving, smiling, looking, wanting, blowing, wishing, chanting, dancing, emotions, thoughts, hand dreams, intoxication, happiness, swimming, playing, shaking, singing and laughing in Li Bai.

Conclusion

• Sentiment/mood index: Du Fu's poetry leans towards depression (lower), while Li Bai's leans towards cheerful and refreshing (higher).
• Du Fu's worldview focuses on the ethical, on tangible senses of actual experiences, and societal issues, "Entering the World: Saint."
• Li Bai's worldview leans towards psychological refraction of "sense" as well as virtual or remote worlds, "Leaving the World: Immortal."
• Language statistics show that they have obvious preferences or biases toward their inner selves and the external world.
• By statistically analyzing specific corpora and analyzing them with reliable and objective data, we can comprehensively understand differences in the artistic styles of Li Bai and Du Fu's poetry.
• The fundamental cause of these differences is their worldview (i.e., their perception of the world's viewpoint).