Hubert Ronald

Appearance

VersoVector ​

VersoVector is a public NLP/MLOps repository for exploring emotional-semantic patterns in poetic and lyrical language.

The project explores whether computational language models can detect emotional, thematic, and semantic relationships in poems in a way that supports human literary interpretation.

It combines two complementary learning approaches:

  • Unsupervised learning: clustering poems by style, tone, topic, or semantic proximity.
  • Supervised learning: classifying poems by emotion, theme, or poetic tone.

The central research question is:

Can a language model perceive the emotion behind a poem, as a human reader does?

What this documentation is for ​

This documentation helps a reader:

  • clone the public repository;
  • prepare the local environment;
  • understand the notebook sequence;
  • understand the model topology;
  • reproduce the main analytical pipeline;
  • inspect generated outputs;
  • understand how model artifacts are packaged;
  • run the local API/frontend foundation when the model bundle is available.

Repository focus ​

The public repository demonstrates:

  • text cleaning and preprocessing;
  • FeatureUnion-based feature representation;
  • CountVectorizer, TfidfVectorizer, and DictVectorizer;
  • supervised multilabel tag prediction;
  • unsupervised similarity, topic modeling, clustering, and projections;
  • integration of supervised and unsupervised outputs;
  • model bundle generation;
  • Python inference abstractions;
  • FastAPI serving foundation;
  • Gradio frontend foundation;
  • Docker-based local services;
  • tests and reproducibility-oriented structure.

Start here if you want to clone the repository, reproduce the notebooks, and understand how the model pipeline is assembled.

Local SetupPrepare Python, dependencies, notebooks, and local tooling.DatasetDownload PoetryFoundationData.csv and place it under data/.Notebook GuideRun the notebooks in the correct order and understand each stage.Model TopologySee how supervised and unsupervised learning branches connect.PipelineFollow the flow from raw poems to integrated outputs.ArchitectureUnderstand repository layers, model bundle, inference, and services.Results GuideLearn how to read predictions, topics, clusters, and visual outputs.Serving & DemoRun the local API/frontend foundation when artifacts are ready.

Source repository ​

Future direction ​

This public repository may later support a hosted product demo, but this documentation focuses on the reproducible public project: how to run it, inspect it, and understand its results.