SeDa: A Unified System for Dataset Discovery and Multi-Entity Augmented Semantic Exploration

Kan Ling; Zhen Qin; Yichi Zhu; Hengrun Zhang; Huiqun Yu; Guisheng Fan

cs.IRMar 8, 2026

SeDa: A Unified System for Dataset Discovery and Multi-Entity Augmented Semantic Exploration

Authors: Kan Ling, Zhen Qin, Yichi Zhu, Hengrun Zhang, Huiqun Yu, Guisheng Fan

Organizations: East China University of Science and Technology, Shanghai, China

Abstract

The continuous expansion of open data platforms and research repositories has led to a fragmented dataset ecosystem, posing significant challenges for cross-source data discovery and interpretation. To address these challenges, we introduce SeDa--a unified framework for dataset discovery, semantic annotation, and multi-entity augmented navigation. SeDa integrates more than 7.6 million datasets from over 200 platforms, spanning governmental, academic, and industrial domains. The framework first performs semantic extraction and standardization to harmonize heterogeneous metadata representations. On this basis, a topic-tagging mechanism constructs an extensible tag graph that supports thematic retrieval and cross-domain association, while a provenance assurance module embedded within the annotation process continuously validates dataset sources and monitors link availability to ensure reliability and traceability. Furthermore, SeDa employs a multi-entity augmented navigation strategy that organizes datasets within a knowledge space of sites, institutions, and enterprises, enabling contextual and provenance-aware exploration beyond traditional search paradigms. Comparative experiments with popular dataset search platforms, such as ChatPD and Google Dataset Search, demonstrate that SeDa achieves superior coverage, timeliness, and traceability. Taken together, SeDa establishes a foundation for trustworthy, semantically enriched, and globally scalable dataset exploration.

Figures & tables

Panel of Figure 1 — Figure 1. Illustration of missing metadata elements in (a) Google Dataset Search and (b) ChatPD. Here, the dataset MS COCO has the URL, while Teapots and 3DShapes does not.

Figure 2. Overall architecture of the proposed system.

Figure 4. Navigation module visualization.

Variant	Topic Relevance		Availability	Exploration
	Win%	Score	Alive%	Datasets (Gain%)
Full (SeDa)	84.40	2.60	99.84	519 (+10.9%)
w/o Topic	10.14	1.67	–	–
w/o Dead-link	–	–	94.68	–
w/o Multi-entity	–	–	–	468

Table 1. Query-centric ablation results.

Evaluator	Preference (%)			Relevance Score (0–3)
	Direct	Topic	Tie	Direct	Topic	$\Delta$
GPT-5.2	9.38	84.38	6.24	1.615	2.619	+1.004
Gemini 3 Pro	10.90	85.20	3.90	1.689	2.607	+0.918
DeepSeek-V3.2	11.70	83.60	4.70	1.629	2.539	+0.910
Qwen3-Max	8.60	84.40	7.00	1.739	2.642	+0.903
Avg.	10.14	84.40	5.46	1.668	2.602	+0.934

Table 2. LLM evaluation breakdown.

Dataset Name	Ours	ChatPD	GDS	Tag	Source
SDSS-V DR19 ( Pallathadka et al., 2025 )	✓	✗	✗	astronomy, stellar observation	github
LOONGBENCH ( Huang et al., 2025 )	✓	✗	✗	reasoning enhancement, synthetic data generation	github
GlobalBuildingAtlas ( Zhu et al., 2025 )	✓	✗	✗	geospatial data, sustainable development	mediaTUM
AthleticsPose ( Suzuki et al., 2025 )	✓	✗	✗	sports data analysis, human pose estimation	github
AgriPotential ( Sakka et al., 2025 )	✓	✗	✗	agricultural potential prediction, remote sensing	zenodo

Table 3. Availability and metadata of representative newly discovered datasets (June–September 2025).

Figure 5. Monthly comparison of newly found datasets and matches in ChatPD and GDS.

Appendix figures & tables8 assets

Supplementary material from the paper’s appendix.

Appendix

Dataset	Our Tags	HuggingFace Task	HuggingFace Tag	Description (excerpt)
CTI5	traditional Chinese instruments, music classification	audio classification	music, art	Recordings of Chinese traditional musical instruments for timbre recognition.
Rico	mobile application, user interface design	others	graphic-design	Large-scale dataset of mobile app screenshots and UI metadata.
TreeSatAI-T	tree species classification, temporal series analysis	image classification	earth-observation, remote-sensing	TreeSatAI time series dataset for tree species classification.
batman2	traditional Chinese medicine, drug discovery	others	biology, drug discovery	Molecular structures derived from traditional Chinese medicine for drug discovery.
clinc_oos	intent classification, out-of-scope query detection	text classification	intent classification	Dataset for out-of-scope intent detection in dialogue systems.

Appendix

Table 4. Comparison between our generated topic tags and HuggingFace annotations.

Field	Record 1	Record 2
dataset_name	Tiny_Imagenet	Tiny ImageNet
dataset_desc	Tiny Imagenet has 200 Classes, each class has 500 traininig images, 50 Validation Images and 50 test images. Label Classes and Bounding Boxes are provided.	Tiny ImageNet contains 100000 images of 200 classes (500 for each class) downsized to 64×64 colored images. Each class has 500 training images, 50 validation images and 50 test images.
dataset_url	https://figshare.com/articles/dataset/Tiny_Imagenet	https://paperswithcode.com/dataset/tiny-imagenet

Appendix

Table 5. Example of duplicate records detected by semantic similarity where explicit identifier matching fails.

Site	$N_{s}$	$S_{1}$	$A_{1}$	$S_{2}$	$A_{2}$
DataCite	2,103,855	2,598	0.89	2,696	0.99
data.europa	1,761,468	2,564	0.88	2,654	1.0
Kaggle	359,561	1,148	0.96	1,199	1.0
HuggingFace	193,227	858	0.85	879	0.98
data.gov	71,424	531	0.86	534	1.0
arXiv	17,743	272	0.90	266	1.0

Appendix

Table 6. Two-round site-level sampling and alive-rate results. (

S_{1}

S_{2}

: sample size;

A_{1}

A_{2}

: alive rates.)

Dataset Discovery

See all 9 papers

A Framework and Prototype for a Navigable Map of Datasets in Engineering Design and Systems Engineering
Mar 16, 2026H. Sinan Bank, Daniel R. HerberDataset Discovery Information Retrieval
Department of Systems Engineering, Colorado State University, Fort Collins, CO 80523
Towards Intelligent Geospatial Data Discovery: a knowledge graph-driven multi-agent framework powered by large language models
Mar 21, 2026Ruixiang Liu, Zhenlong Li, Ali Khosravi KazaziGraph-Based Retrieval LLM Agents
Geoinformation and Big Data Research Laboratory, Department of Geography, The Pennsylvania State University, University Park, PA, USA
Scientific Data Skills: Enabling Agent-Ready Scientific Data Services at Scale
Aug 20, 2026Xiaohan Huang, Qingqing Long, Xiaolei Du +8Knowledge Representation LLM Agent Skill Retrieval
Computer Network Information Center, Chinese Academy of Sciences University of the Chinese Academy of Sciences Beijing, China

Mar 16, 2026J/K move · Enter open · S save

A Framework and Prototype for a Navigable Map of Datasets in Engineering Design and Systems Engineering

H. Sinan Bank, Daniel R. Herber

Department of Systems Engineering, Colorado State University, Fort Collins, CO 80523

Dataset Discovery Information Retrieval

Open paper AskPDF arXiv

The proliferation of data across the system lifecycle presents both a significant opportunity and a challenge for Engineering Design and Systems Engineering (EDSE). While this "digital thread" has the potential to drive innovation, the fragmented and inaccessible nature of existing datasets hinders method validation, limits reproducibility, and slows research progress. Unlike fields such as computer vision and natural language processing, which benefit from established benchmark ecosystems, engineering design research often relies on small, proprietary, or ad-hoc datasets. This paper addresses this challenge by proposing a systematic framework for a "Map of Datasets in EDSE." The framework is built upon a multi-dimensional taxonomy designed to classify engineering datasets by domain, lifecycle stage, data type, and format, enabling faceted discovery. An architecture for an interactive discovery tool is detailed and demonstrated through a working prototype, employing a knowledge graph data model to capture rich semantic relationships between datasets, tools, and publications. An analysis of the current data landscape reveals underrepresented areas ("data deserts") in early-stage design and system architecture, as well as relatively well-represented areas ("data oases") in predictive maintenance and autonomous systems. The paper identifies key challenges in curation and sustainability and proposes mitigation strategies, laying the groundwork for a dynamic, community-driven resource to accelerate data-centric engineering research.

Abstract

Dataset Discovery

Explore similar work

A Framework and Prototype for a Navigable Map of Datasets in Engineering Design and Systems Engineering

Towards Intelligent Geospatial Data Discovery: a knowledge graph-driven multi-agent framework powered by large language models

Scientific Data Skills: Enabling Agent-Ready Scientific Data Services at Scale