HSC-Bench

Leaderboard

Centralized benchmark results for service recommendation and service composition, with static filters for the first release.

How to read the leaderboard

The first release uses static CSV-backed tables. Results should specify dataset, model type, code availability, official reproduction status, and whether the result follows the unified split or protocol.

Service Recommendation Service Composition

Service Recommendation

Dataset Model Type

Model	Dataset	Type	K	P@5	P@10	R@5	R@10	NDCG@5	NDCG@10	MRR	Code	Paper
SRLCF	HSC+	Benchmark Model	5	TBD	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper
MTFM	HSC+	Neural	10	TBD	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper
GSAT	HSC+	Graph-based	10	TBD	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper
GSL-Mash	ProgrammableWeb	Graph-based	10	TBD	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper
Frequency	HSC+	Traditional	10	TBD	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper
LLM-based reranking	HSC+	LLM-based	10	TBD	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper

Service Composition

Dataset Model Type

Model	Dataset	Type	Utility ↑	RT ↓	Cost ↓	Throughput ↑	Availability ↑	Reliability ↑	Code	Paper
GNNPN-SC	HSC+	Learning-based	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper
SDFGA	HSC+	Optimization-based	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper
DAAGA	HSC+	Optimization-based	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper
GA	QWS	Optimization-based	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper
LLM Planner	HSC+	LLM-based	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper
Multi-agent pipeline	HSC+	LLM-based	TBD	TBD	TBD	TBD	TBD	TBD	Code	Paper

Submission protocol

Use the official dataset version and split.
Run the published evaluation script with fixed random seed and configuration.
Add a row to the corresponding CSV file.
Submit a pull request with paper link, code link, configuration, logs, and hardware information.