Topic Modeling using Latent Dirichlet Allocation (LDA)

This repository presents a simple application of the Latent Dirichlet Allocation (LDA) model for topic modeling. The project consists of applying the LDA model to the document collection of the GT-8 ENANCIB (Information and Technology of the National Meeting of Research in Information Science).

Collection

• The corpus consists of 273 documents (full articles and extended abstracts between 2012 and 2019).

Requirements

• Python3
• python3.8-venv

Exploratory Data Analysis

• Wordcloud

Data pre-processing

• Conversion of terms
• Stopwords configuration
• Bigram e trigram models
• Tokenization

What can still be done

• Removal of words less than 3 characters
• Removal of irrelevant words
• Stemming e Lemmatization

How to run

• Start the project by preparing the environment through the command:

$ source prepare-env.sh

• Now you can use the notebook TopicModel_LDA in your prefer code editor.