MA4DIV

This is the code of MA4DIV which is a reinforcement learning approach for search result diversification. The framework of this project is based on https://github.com/hijkzzz/pymarl2.

Run an Experiment

python main.py --config=qmix --env-config=search_engine

Hyperparameters

Some hyperparameters can be edited in the file of:

./config/envs/search_engine.yaml

Data Loading Path

The data path can be edited in the file of:

./runners/episode_runner.py

Data Case

The indexes of queries and documents in the DU-DIV are shown in:

./data/baidu/query_doc_subtopics_doc15_2.json

We will release the whole DU-DIV dataset in the future. The whole dataset includes the features of 4473 queries and 67095 documents with $1024$-dimensions.