Supervised vs Unsupervised Latent DirichletAllocation: topic detection in lyrics.

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Topic modeling is a type of statistical modeling for discovering the abstract ``topics'' that occur in a collection of documents. Latent Dirichlet Allocation (LDA) is an example of topic model and is used to classify text in a document to a particular topic. It builds a fixed number of topics starting from words in each document modeled according to a Dirichlet distribution. In this work we are going to apply LDA to a set of songs from four famous Italian songwriters and split them into topics. This work studies the use of themes in lyrics using statistical analysis to detect topics. Aim of the work is to underline the main limits of the standard unsupervised LDA and to propose a supervised extension based on the Correspondence Analysis (CA) association theory.
Original languageEnglish
Title of host publicationBook of short papers - SIS 2020
Number of pages6
Publication statusPublished - 2020


Dive into the research topics of 'Supervised vs Unsupervised Latent DirichletAllocation: topic detection in lyrics.'. Together they form a unique fingerprint.

Cite this