To perform aspect-based sentiment analysis we first need to extract aspects or say topics short text like tweets. Will techniques like LDA give up to the mark performance for this kind of topic extraction?
LDA, as a supervised learning classifier, will use the extracted features from the text to make a *specific* decision about the examined data. The performance of LDA depends on several factors; hence, to evaluate it, you may look at, e.g., accuracy, confusion, matrix, etc.
HTH.
Dr. Samer Sarsam
Salah Eddine Labiad It depends as far my experiment, LDA will perform good as doesn't depend on length of text like it will treat tweets as document.
LDA doesn't perform well in short text especially on tweets. unless use LDA extension TLDA to overcome the problem of sparsity.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts