APEX: daptive Variable-wise Parallel Execution for Worst-Case Optimal Joins on Graph Queries

APEX is a graph query engine for RDF data based on worst-case optimal joins (WCOJ). It supports variable-wise parallel execution and dynamic variable ordering, and is designed to efficiently process complex graph queries on large-scale knowledge graphs.

---

Features

• Build RDF databases with trie-based indexes
• Execute graph queries using worst-case optimal joins
• Train and test variable ordering models
• Multi-threaded query execution

---

Build

1. Clone this project

git clone git@github.com:MKMaS-GUET/APEX.git
git submodule update --init

2. Build this project

cmake -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build

Or use the build.sh script to build this project directly

./scripts/build.sh

---

Usage

Build Database

./apex build -d <DATABASE_PATH> -f <RDF_DATA_FILE>

Query Database

./apex query -d <DATABASE_PATH> -f <QUERY_FILE> -t <NUM_THREADS>

Train Model

./apex train -d <DATABASE_NAME> -f <QUERY_FILE> -t <NUM_THREADS>

Test Model

./apex test -d <DATABASE_NAME> -f <QUERY_FILE> -t <NUM_THREADS>

---

Options

Option	Description
-d, --database	Database path or name
-f, --file	RDF data file or query file
-t, --threads	Number of threads

---

Example

./apex build -d mydb -f data.nt
./apex query -d mydb -f queries.txt -t 16

---

Notes

• If the database path does not contain /, it will be stored under DB_DATA_ARCHIVE/ by default.
• The default number of threads is 32.