Bimodal Music Subject Classification via Context-Dependent Language Models

Choi, Kahyun

doi:10.1007/978-3-030-71292-1_7

Cited by 1 publication

(4 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this paper, we employ BERT and SVM for multilabel-multimodal poem classification, extending their previous application in single-label classification tasks (Choi, 2021. This key structural modification reflects the fact that real-world poems can contain multiple themes simultaneously.…”

Section: Classification Setupmentioning

confidence: 96%

“…To bridge this gap, we propose a poem topic classification algorithm utilizing a pretrained large language model, building upon the work of Lou et al (2015) and Choi (2021). Lou et al (2015) introduced the multilabel poem topic classification task and explored its potential using support vector machines (SVM) on Poetry Foundation's poem data.…”

Section: Introductionmentioning

confidence: 99%

“…Lou et al (2015) introduced the multilabel poem topic classification task and explored its potential using support vector machines (SVM) on Poetry Foundation's poem data. We extend this study by examining the applicability of pretrained large language models, inspired by findings from another work on song lyrics (Choi, 2021). Choi's work focused on the song lyrics, which were assumed to consist of hard‐to‐interpret choices of words, making computational analysis difficult.…”

Section: Introductionmentioning

confidence: 99%

“…From both the AI model and human reader's perspectives, certain types of auxiliary data are easier to interpret due to their more straightforward language. In previous research on song lyrics, for example, music listeners' comments successfully assisted AI models in understanding themes within song lyrics (Choi et al, 2016; Choi, 2021). In this work, we propose to use a different kind of auxiliary information, such as author's notes, by assuming that they can demystify symbolic text written by authors themselves.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Computational Thematic Analysis of Poetry via Bimodal Large Language Models

Choi

2023

Proceedings of the Association for Information Science and Tech

Self Cite

View full text Add to dashboard Cite

This article proposes a multilabel poem topic classification algorithm utilizing large language models and auxiliary data to address the lack of diverse metadata in digital poetry libraries. The study examines the potential of context‐dependent language models, specifically bidirectional encoder representations from transformers (BERT), for understanding poetic words and utilizing auxiliary data, such as author's notes, in supplementing poetry text. The experimental results demonstrate that the BERT‐based model outperforms the traditional support vector machine‐based model across all input types and datasets. We also show that incorporating notes as an additional input improves the performance of the poem‐only model. Overall, the study suggests pretrained context‐dependent language models and auxiliary data have potential to enhance the accessibility of various poems within collections. This research can eventually assist in promoting the discovery of underrepresented poems in digital libraries, even if they lack associated metadata, thus enhancing the understanding and appreciation of the literary form.

show abstract

Section: Classification Setupmentioning

confidence: 96%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations