diff --git a/README.md b/README.md
index 10b269ef488b437d60cd2f8a11c44686c7df1330..c242009ccfa7c57683084b31c0111c3e67e47de0 100644
--- a/README.md
+++ b/README.md
@@ -51,7 +51,7 @@ and compile the framework.
 In case of problems with the automatic setup, you can prepare the environment manually:
 
 1. Download [Apache Flink 1.14.0](https://archive.apache.org/dist/flink/flink-1.14.0/flink-1.14.0-bin-scala_2.11.tgz) and decompress in `repo_dir`
-3. Get the input datasets, either using `./get-datasets.sh` or by manually downloading from [here](https://drive.google.com/u/0/uc?id=1uNOUlCoa9CfH7WCxe3nSsCwQVVjf9teB) and decompressing in `repo_dir/data/input`
+3. Get the input datasets, either using `./get-datasets.sh` or by manually downloading from [here](https://zenodo.org/records/14007044) and decompressing in `repo_dir/data/input`
 4. Compile the two experiment jars, from repo dir: `mvn -f helper_pom.xml clean package && mv target/helper*.jar jars; mvn clean package; mvn install`
 2. Run `./init-configs.sh` to use the Flink configurations used in our experiments