{"id":1982,"date":"2025-05-04T12:46:02","date_gmt":"2025-05-04T16:46:02","guid":{"rendered":"https:\/\/students.bowdoin.edu\/bowdoin-science-journal\/?p=1982"},"modified":"2025-05-04T12:46:02","modified_gmt":"2025-05-04T16:46:02","slug":"unsupervised-thematic-clustering-for-genre-classification-in-literary-texts","status":"publish","type":"post","link":"https:\/\/students.bowdoin.edu\/bowdoin-science-journal\/csci-tech\/unsupervised-thematic-clustering-for-genre-classification-in-literary-texts\/","title":{"rendered":"Unsupervised Thematic Clustering for Genre Classification in Literary Texts"},"content":{"rendered":"
\"Book
(Chapterly 2022)<\/figcaption><\/figure>\n

Summary<\/span><\/h1>\n

In the last decade, computational literary studies have expanded, yet computational thematics remains less explored than areas like stylometry, which focuses on identifying stylistic similarities between texts. A 2024 study by researchers from the Max Planck Institute and the Polish Academy of Sciences investigated the most effective computational methods for measuring thematic similarity in literary texts, aiming to improve automated genre clustering.<\/span><\/p>\n

Key Findings and Assumptions<\/span><\/h2>\n
    \n
  • Key Assumptions:<\/b>\u00a0<\/span>\n
      \n
    • Text pre-processing to emphasize thematic content over stylistic features could improve genre clustering.\u00a0<\/span><\/li>\n
    • Unsupervised clustering would offer a more scalable and objective approach to genre categorization than manual tagging by humans.<\/span><\/li>\n
    • Four genres were selected (detective, fantasy, romance, science fiction) for their similar level of broad qualities.<\/span><\/li>\n
    • If the genres are truly distinct in terms of themes, computers should be able to separate them into clusters.<\/span><\/li>\n<\/ul>\n<\/li>\n
    • Best Performance:<\/b> The best algorithms were 66-70% accurate at grouping books by genre. Thus showing unsupervised genre clustering is feasible despite the complexity of literary texts.<\/span><\/li>\n
    • Text Pre-Processing:<\/b> Medium and strong levels of text pre-processing significantly improved clustering, while weak pre-processing performed poorly.<\/span><\/li>\n
    • Which methods worked best:<\/b> Doc2vec, a method that captures word meaning and context, performed the best overall, followed by LDA (Latent Dirichlet Allocation), which finds major topics in texts. Even the simpler bag-of-words method, which just counts how often words appear, gave solid results.<\/span><\/li>\n
    • Best way to compare genres:<\/b> Jensen-Shannon divergence, which compares probability distributions, was the most effective metric, while simpler metrics like Euclidean distance performed poorly for genre clustering.<\/span><\/li>\n<\/ul>\n

      Methodology\u00a0<\/span><\/h1>\n

      Sample Selection<\/span><\/h2>\n

      The researchers selected canonical books from each of the four genres, ensuring they were from the same time period to control for language consistency.<\/span><\/p>\n

      Sample Pre-Processing and Analysis\u00a0<\/span><\/h2>\n

      The researchers analyzed all 291 combinations of the techniques in each of the three stages: text pre-processing, feature extraction, and measuring text similarity.\u00a0<\/span><\/p>\n

      Stage 1: Different Levels of Text Pre-Processing\u00a0\u00a0<\/b><\/p>\n