goal: implement a news recommendation algorithm, that predicts user engagement with articles.
assigned model: gru - gated recurrent unit
general competition links:
- assignment: https://tuwel.tuwien.ac.at/mod/page/view.php?id=2281418
- gitlab: https://gitlab.tuwien.ac.at/recsys-laboratory/teaching/24ss-recsys-lecture/Group_26
- acm recsys challenge: https://www.recsyschallenge.com/2024/
- acm conference: https://recsys.acm.org/recsys24/challenge/
- codabench: https://www.codabench.org/
development links:
-
dataset:
- description: https://recsys.eb.dk/dataset/
-
ebnerd-benchmark:
-
recommenders:
- quickstart: https://github.com/ebanalyse/ebnerd-benchmark/blob/main/examples/00_quick_start/dataset_ebnerd.ipynb
- LSTUR implementation: https://github.com/recommenders-team/recommenders/blob/main/examples/00_quick_start/lstur_MIND.ipynb 👈
- code: https://github.com/recommenders-team/recommenders/blob/main/recommenders/models/deeprec/models/sequential/gru.py
- docs: https://recommenders-team.github.io/recommenders/models.html#gru
rnn theory:
- https://colah.github.io/posts/2015-08-Understanding-LSTMs/
- https://colah.github.io/posts/2015-09-NN-Types-FP/
- https://karpathy.github.io/2015/05/21/rnn-effectiveness/
- statquest: https://www.youtube.com/watch?v=zxagGtF9MeU&list=PLblh5JKOoLUIxGDQs4LFFD--41Vzf-ME1
steps:
- register group and algorithm on tuwel, register organization on "codabench" ✅
- download dataset ✅
- implement 3 algorithms:
- 1 - baseline model from the repository "NRMS on EB-NeRD" from "ebnerd-benchmark" ✅
- 2 - "GRU" from "recommenders" repository (it's the core of LSTUR) ✅
- 3 - some algorithm of choice ❌ → not implemented yet
- improve: check out "beyond metrics" section in the "ebnerd-benchmark" repository
- write report pdf
- submit to "codabench"
- submit to gitlab, add "final" tag
- upload project report pdf