박물관 전시 경험 분석을 위한 다층적 접근 방법론: 인스타그램 데이터에 대한 토픽 모델링과 이미지 분석

초록

This study represents a convergent research methodology integrating arts management and technology, analyzing visitor experiences of 《Blooming Hwarot: Bridal Robes of the Joseon Royal Court》through topic modeling and image recognition. Text mining of Instagram posts with TF-IDF and LDA was applied to identify both objective information (exhibition title, theme, location) and subjective responses, including storytelling on restoration, curatorial elements, and expressions of admiration and beauty. The LDA results show that visitors emphasize the cultural value and conservation of the hwarot. Image analysis, using Google Cloud Vision API and Word2Vec, extracts 39,335 labels, categorized into eleven thematic groups. Fashion groups have the highest label frequencies, while Exhibition accounts for the largest proportion across the dataset.

키워드