Background For Topic Modeling

Background for Topic Modeling

Created: 2022-09-12 10:35
#note

This section of the paper is going to be divided in two sub-sections:

1. Terms and concepts in Topic modeling;
2. Literature review

Terms and concepts

Concepts:

• Topic modelling;
• Document;
• Word;
• Vocabulary;
• Topic;
• Evaluate Clustering

Terms:

• BoW -> efficient but does not consider order or semantic meaning;
• TF-IDF;
• Embeddings;
• Transformers;

Describe concepts in TM and from the definition of 'word' pass to the various ways in which a word is represented (BoW, TF-IDF and embeddings). Then discuss the different methods to compute embeddings (word2vec, transformers etc), the pros and cons of the main methods (bonus: write about approaches to deal with long texts). At the end, present the evaluation metrics used.

Literature review

• History;
• Most important methods and papers in the past;
• Fields of application;
• Tourism

Start from a bit of historical stepstones for TM, the most used methods in the past and the fields in which TM was used. End this section with some references to papers about TM for tourism.

References

1. Wikipedia
2. Review of topic modeling methods
3. Review of topic modeling approaches for short texts

Code

Tags

#paper #topicmodeling