COMPARATIVE STUDY OF RECURRENT NEURAL NETWORK-BASED ALGORITHMS AND ATTENTION MECHANISM IN SESSION-BASED RECOMMENDATION SYSTEMS
Recommendation system is one of the technologies that is continuously being developed for utilization in various domains. Recommendation systems are useful for enhancing user convenience in making decisions. Session-based recommendation system is an emerging paradigm that focuses on learning user...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/84265 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Recommendation system is one of the technologies that is continuously being
developed for utilization in various domains. Recommendation systems are useful
for enhancing user convenience in making decisions. Session-based
recommendation system is an emerging paradigm that focuses on learning users'
short-term and dynamic preferences. The use of recurrent neural networks (RNN)
has become a popular algorithm choice due to its ability to learn sequential data,
and attention mechanism is increasingly being utilized alongside RNNs to
recognize user intentions within a session. Therefore, the performance of two RNN
variants, namely LSTM and GRU, will be analyzed when combined with attention
mechanism to perform session-based recommendation tasks. The two developed
models will be compared based on recall and mean reciprocal rank (MRR) metrics.
Additionally, the performance of both models in handling various types of session
data based on session length will also be compared. |
|---|