Topic Detection Based on Movie Reviews

Topic Detection Analysis - Movie Reviews

Topic detection extracts relevant information elements from unstructured text documents and groups them to define a number of topics. This workflow illustrates how to perform a topic detection analysis on movie reviews.

Task

Perform a topic detection in IMDb reviews.

Data Reading

Read IMDb reviews from a CSV file.

The file is located in TheData/SocialMedia

Preprocessing

Classic pre-processing of documents: Punctuation Erasure, Number Filter, N Chars Filter, Stop Word Filter, Case Converter

Double click the metanode to see the subworkflow

Topic Detection

Build a list of topics of the pre-processed documents using the Topic-Extractor (Parallel LDA) node. Use 4 words for each topic and 8 topics.

Grouping

The GroupBy node concatenates the keywords for the identified topics.

Try this:

1. Go to the configuration window of the Topic Extractor (Parallel LDA)

2. Try to change the number of words and topics that you would like to detect in the document.