Sie können Operatoren mit Ihrer Suchanfrage kombinieren, um diese noch präziser einzugrenzen. Klicken Sie auf den Suchoperator, um eine Erklärung seiner Funktionsweise anzuzeigen.
Findet Dokumente, in denen beide Begriffe in beliebiger Reihenfolge innerhalb von maximal n Worten zueinander stehen. Empfehlung: Wählen Sie zwischen 15 und 30 als maximale Wortanzahl (z.B. NEAR(hybrid, antrieb, 20)).
Findet Dokumente, in denen der Begriff in Wortvarianten vorkommt, wobei diese VOR, HINTER oder VOR und HINTER dem Suchbegriff anschließen können (z.B., leichtbau*, *leichtbau, *leichtbau*).
Dieses Kapitel geht auf das transformative Potenzial künstlich unterstützter Dienstleistungen im Bereich der Transportforschung ein, wobei ein besonderer Schwerpunkt auf der Gewinnung und Erstellung von Inhalten liegt. Der Text untersucht die Entwicklung und Umsetzung eines domänenspezifischen Wissensdiagramms (Knowledge Graph, KG), der Daten aus verschiedenen Quellen integriert, um die wissenschaftliche Qualität zu verbessern und Forschungsaktivitäten über nationale Grenzen hinweg auszutauschen. Zu den Schlüsselthemen zählen die Struktur und Anwendung von Wissensgrafiken, ihre Rolle bei der Informationsgewinnung und die Integration von KI-gestützten Diensten für Smart Browsing und Datenanalyse. Das Kapitel stellt drei detaillierte Szenarien vor, die den praktischen Nutzen dieser Dienstleistungen veranschaulichen, darunter Benchmark-Reproduzierbarkeitsberichte für Algorithmen, Benchmark-Datensatzberichte zu Optimierungszwecken und Generalisierungsberichte für Forschungspublikationen. Die Ergebnisse unterstreichen die Effektivität künstlich unterstützter Dienste bei der Förderung offener und nachhaltiger wissenschaftlicher Kommunikation im Verkehrsforschungssektor. Das Kapitel schließt mit einer Diskussion über zukünftige Forschungsrichtungen, in der die Bedeutung ethischer und rechtlicher Überlegungen bei der Einführung künstlicher Intelligenz und Open-Access-Initiativen betont wird.
KI-Generiert
Diese Zusammenfassung des Fachinhalts wurde mit Hilfe von KI generiert.
Abstract
This paper presents the gateway of the domain-specific Knowledge Graph (KG), which was built based on related data sources and by exploiting the OpenAIRE ecosystem and EOSC services. These enabled it to offer services for integrated KG smart browsing based on impact and reproducibility using AI by also serving several categories of stakeholders. Following the current trends and stakeholders’ needs, the areas of highest interest were identified and gaps in data and knowledge were also detected. On top of that knowledge space, transport research-inspired information retrieval scenarios were implemented by tuning the use of individual AI-assisted services or combinations of them. More specifically, scenarios were built around different use case directions including reproducibility/reusability reports for publications and/or datasets in the transport research sector which were automatically identified and they included information about the ease to reproduce/reuse them and the extent to which the work has already been reproduced by meta-analysis studies.
1 Introduction
Transportation research sector involves different modes of transport (i.e., road, rail, water, air, cross-modal) and different types of mobility (i.e., passenger and freight) producing innovation uptakes that could enable users and societies to be better informed and make safer, more coordinated and ‘smarter’ use of transport systems. More specifically, information processing is combined with engineering studies, technologies and human sciences producing research data span multiple domains (e.g., algorithms, software Proof of Concepts (PoCs), methods for technological impact, user acceptance, inputs to draft standards and regulations of newly introduced technologies, etc.). In addition, several domain-specific knowledge bases (i.e., TRID, ERTRAC, ERTICO, ECTRI, IRU, WEGEMT, EASA, ERRAC) are exploited to enhance scientific quality, identify remarkable knowledge and share research activities across national boundaries.
In this context, a domain-specific Knowledge Graph (KG) developed by relying on related data sources and exploiting the OpenAIRE ecosystem and EOSC services. These enabled it to offer services for integrated KG smart browsing based on impact and reproducibility using AI by also serving several categories of stakeholders (i.e., researchers, developers, regulators). The areas of highest interest were identified and gaps in data and knowledge were also detected. On top of that knowledge space, transport research-inspired information retrieval scenarios were implemented by tuning the use of individual AI-assisted services or combinations of them. More specifically, scenarios were built around different use case directions including reproducibility/reusability re-ports for publications and/or datasets in the transport research sector which were automatically identified, and they included information about the ease to reproduce/reuse them and the extent to which the work has already been reproduced by meta-analysis studies.
Anzeige
The remainder of this paper is organized as follows: Sect. 2 provides an overview of the key developments and trends in the field of knowledge graphs up to the present day. The followed domain-specific Knowledge Graph in the context of transport research and the derived results of its implementation are presented in Sect. 3. Finally, in Sect. 4 conclusions are drawn and pointers for future research are provided.
2 Literature Review
Knowledge graphs have emerged as a fundamental tool for organizing and representing structured knowledge in a graph-based format (Hogan et al., 2021). They have gained significant attention in recent years due to their wide-ranging applications in various domains, including natural language processing, data integration, semantic web, recommendation systems, and artificial intelligence (Chen et al., 2020). The concept of knowledge graphs can be traced back to early AI research, but it gained substantial traction with the advent of the Semantic Web initiative (Chaudhri et al., 2022). Researchers recognized the need for a standardized, machine-readable representation of knowledge to enable machines to understand and reason about information on the web.
The key components of Knowledge Graphs (KG) involve nodes and edges with semantics underlying, and they could be represented as Resource Description Framework (RDF) or Web Ontology Language (OWL). The KGs consist of nodes representing entities (e.g., people, places, concepts) and edges representing relationships between these entities (Hamilton et al., 2018). This graph-based structure allows for the representation of complex relationships and hierarchical information. In addition, the RDF and the OWL (Gutiérrez and Sequeda, 2021) are commonly used formalisms for representing knowledge in a structured and machine-readable way. RDF provides a simple, flexible format for expressing subject-predicate-object, while OWL allows for more complex ontology modeling.
The application of KGs varies as they could be used in (i) semantic search (Li, 2017) allowing users to retrieve relevant information and answer complex queries, (ii) in recommendation systems (Dongliang et al., 2022) to help in understanding user preferences and item attributes, (iii) in question answering (Chakraborty et al., 2021) for extracting relevant information from vast knowledge bases to provide concise answers to user queries, and (iv) in natural language processing (He et al., 2022) for entity recognition, sentiment analysis, and text summarization by incorporating semantic context.
Anzeige
In recent years, several techniques have been developed to learn continuous vector representations of entities and relationships in knowledge graphs, enabling better reasoning and link prediction (Ali et al., 2021). Moreover, large-scale KGs (such as Wikidata and Google Graph) created and offer vast, interconnected knowledge bases covering a wide range of domains. KGs also integrated with Graph Neural Networks (GNNs) to improve the performance in various applications, such as entity linking, node classification, and recommendation (Park et al., 2019). Finally, KGs are used to provide explanations for AI decision-making processes, increasing transparency and trust in AI systems (Lecue, 2020).
3 Knowledge Graph in Transport Research
3.1 Knowledge Graph Structure
The KG consist of five entities, i.e., 1) Results that represent the outcomes of research activities, 2) Data Sources that are the sources from which the metadata of graph objects are collected, 3) Organizations that correspond to companies, research institutions, and public authorities involved in projects, 4) Projects that are research project, and 5) Communities that are groups of people with a common research intent, such as the Transport Research Community. Furthermore, the OpenAIRE research graph is utilized to develop various services offered to three distinct categories of organizations within the Transport Research Community. These categories are: 1) Content Providers, 2) Research Organizations, and 3) Public Authorities.
Regarding Content Providers, OpenAIRE offers validator services to automatically verify compliance of exported records with set guidelines. Additionally, full-text mining algorithms utilize the full-texts of Open Access publications to enhance metadata records by linking them to projects, publications, datasets, software, organizations, research infrastructures, and terms from standard classification schemes. For Research Organizations, the services aim to support the adoption of Open Science publishing practices and monitor the implementation of Open Science practices by their researchers. In cases where an official Open Access repository is unavailable, any organization can easily create a dedicated community and encourage affiliated researchers to deposit their research outputs there. Within the user interface, users can access content reports and project lists. The content report provides information about the research outcomes associated with the organizations, while the project list outlines the projects the organization is involved in and their respective research outcomes. Lastly, Public Authorities, such as national and international research funding organizations, can access services to monitor the impact of their funding within the Transport research community.
The gateway for the Transport Community is provided in the https://beopen.openaire.eu/ focusing on promoting territorial and cross border cooperation and contributing to the optimization of open science in transport research. Integrating the described services and leveraging the OpenAIRE Graph, the Transportation Research community benefits from enhanced publication deposits and tools for in-depth exploration.
3.2 Results
The studied scenarios of transport research sector following the aforementioned gateway of the OpenAIRE, and the derived results are presented below.
(i) The first scenario studies the use case of researchers that are interested in replicating experiments and verify algorithm’s performances on benchmark datasets utilizing open access algorithms/software. Therefore, the developed KG offers a benchmark reproducibility report of algorithms/software (Fig. 1) that addresses complex logistics problems by specifying “open access” in the Access category and using as keywords “optimization”, “algorithms” and “logistics”. Further results with closed access are also found and could be presented by altering our research.
(ii) The second scenario studies the use case of datasets that used for optimization purposes in the logistics sector. The developed KG offers a benchmark datasets report (Fig. 2) that used to assess algorithms performance by utilizing the keyword “optimization”. This keyword is used by the KG to deliver relevant research data sets as it is specified in the Type category selecting “research data”. The results are referring only to the open ones as “open access” was selected in the Access category.
(iii) The third scenario studies the use case of generalization, retrieving information about research publications that study “automation” in different transportation domains. The developed KG offers a generalization report of publications (Fig. 3) including information on studies and experiments, such as how an algorithm’s performance varies with problem conditions or complexity. The keyword “automation” is utilized for searching the KG and open results are presented as “open access” is selected in the Access category.
The developed AI-assisted services could be further enhanced by allowing algorithms comparisons and reproducibility to demonstrate improvements of new algorithms. This should include the code for the baseline algorithms, configuration settings, and clear instructions for conducting fair comparisons in order to enable others to verify the reported performance gains. In addition, parameter tuning is another future step for improvement. Optimization algorithms often have hyperparameters that significantly impact their performance. Thus, reproducibility reports that can outline the process of hyperparameter tuning, including the range of values tested, the criteria for selecting the best configuration, and the scripts used for this purpose could contribute to the advancement of knowledge and the development of more robust optimization algorithms.
4 Conclusions and Future Research
AI-assisted services in the transport research sector introduced to promote science and foster collaboration among research community and other relevant stakeholders. Towards this direction, this paper presents and analyses the services of the gateway for the Transport Community that exploits OpenAIRE Graph. According to the derived results, an open and sustainable scholarly communication is established monitoring and providing all research outcomes in the transport research sector. The presented scenarios and future steps reflect a commitment to advancing research in the sector and optimizing algorithm performance, with a strong emphasis on open-access resources and reproducibility. Consideration is also given on future research and directions are drawn. Ethical and legal considerations should be examined related to AI-assisted services and open access initiatives by exploring issues such as data privacy, intellectual property, and responsible AI deployment in scholarly communication.
Acknowledgements
This project has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement No 101058573. The results in this paper reflect only the authors' view.
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Li, Y.: Research and analysis of semantic search technology based on knowledge graph. In IEEE International Conference on Computational Science and Engineering (CSE) and IEEE International Conference on Embedded and Ubiquitous Computing (EUC), vol. 1, pp. 887–890. IEEE (2017)
Dongliang, Z., Yi, W., Zichen, W.: Review of recommendation systems based on knowledge graph. Data Anal. Knowl. Disc. 5(12), 1–13 (2022)
Chakraborty, N., Lukovnikov, D., Maheshwari, G., Trivedi, P., Lehmann, J., Fischer, A.: Introduction to neural network-based question answering over knowledge graphs. Wiley Interdisc. Rev. Data Min. Knowl. Disc. 11(3), e1389 (2021)CrossRef
He, Q., Yang, J., Shi, B.: Constructing knowledge graph for social networks in a deep and holistic way. In: Companion Proceedings of the Web Conference 2020, pp. 307–308, IEEE (2022)
Ali, M., et al.: PyKEEN 1.0: a python library for training and evaluating knowledge graph embeddings. J. Mach. Learn. Res. 22(1), 3723–3728 (2021)
Park, N., Kan, A., Dong, X. L., Zhao, T., Faloutsos, C.: Estimating node importance in knowledge graphs using graph neural networks. In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp. 596–606 (2019)
Lecue, F.: On the role of knowledge graphs in explainable AI. Semant. Web 11(1), 41–51 (2020)
