Sieve is a learning-enhanced data-skipping index designed to effectively filter out irrelevant data blocks for modern data analytics services. By capturing patterns in the block distribution with piece-wise linear functions, Sieve is able to trade off storage consumption and data filter efficiency.

The experiment dataset can be obtained from the remote dataset of cloudlab.

Cloudlab urn: "urn:publicid:IDN+clemson.cloudlab.us:ccjs-pg0+ltdataset+datasetinclem".

For more details about Cloudlab urn, see http://docs.cloudlab.us/advanced-storage.html#%28part._remote-storage%29

1. StoreSales: https://trino.io/docs/current/connector/tpcds.html
2. Wikipedia: https://dumps.wikimedia.org/other/pagecounts-raw/2016/2016-07/
3. Maps: https://github.com/awslabs/open-data-docs/tree/main/docs/osm-pds

• Data preprocessing(pulling raw data, transforming) can be referred to scripts in ./Sieve/dataset.

Place prepared/generated datasets under path: ./Sieve/dataset/StoreSales, ./Sieve/dataset/Wikipedia, ./Sieve/dataset/Maps.

Ubuntu 18.04

python3.8

```
// for common
sudo apt update
sudo apt upgrade
apt install --upgrade pip

// for data processing
pip install pyarrow==8.0.0
pip install datasketch
pip install pybloom_live

// for cuckoo index
sudo apt-get install pkg-config zip g++ zlib1g-dev unzip python
wget https://github.com/bazelbuild/bazel/releases/download/4.2.3/bazel-4.2.3-installer-linux-x86_64.sh
chmod +x bazel-4.2.3-installer-linux-x86_64.sh
./bazel-4.2.3-installer-linux-x86_64.sh --user
export PATH="$PATH:$HOME/bin"
// cuckoo-index source code
cd ./Sieve
git clone https://github.com/LiuJiazhen1999/cuckoo-index -b change

```

python index/simulator/index_compare/indexcompare_prelog.py.

python index/simulator/insert/insertTest.py.

python index/simulator/blk_scale/diff_blksize.py.

1. Dense data: python index/simulator/worsecasedata/dense_worst_case.py
2. Sparse data: python index/simulator/worsecasedata/sparse_worst_case.py