Support of scientific workflows by semantic technology gains increasing interest in recent years. Huge efforts are put on providing structured, standard-based meta data and on machine based qualitative analysis of unstructured content of scholarly papers. This helps researchers to stay oriented in an ever growing and gaining complexity field. Semantic technologies have also the potential to support the in-depth involvement in scholarly papers, like practiced in research seminars. The paper reports on the preliminary results of an undertaking to support the collaborative documentation and reuse of qualitative analysis of scholarly papers in an information systems research group. A vocabulary is developed and openly provided. The system is implemented on the base of OntoWiki and can be accessed openly.
Research groups form the smallest, often informal social entity in the scientific system. Their performance and their cohesion are mainly based on shared scientific interests and a common, high level of expertise in the research field. Even if this research field is narrowly specified, it remains a great challenge to keep in view the state of knowledge. Beyond the awareness of other research groups and influential researchers in the field, a qualitative expert analysis should focus on research questions, on methods applied to them, as well as on research findings and their critical disputation. Undoubtedly, regular scientific seminars are a traditional and effective instrument for this, since they create a collective realm of experience and discussion.
The small, informal research group Business Modeling and Knowledge Engineering (BMaKE) at the Brandenburg University of Applied Sciences has established such a seminar recently. This group is anchored in the program of information systems. While the selection of the papers to be discussed and the structure to be used in expert analysis were quickly agreed, the form of the knowledge base to be created for storing the analysis results led immediately to the following research question: • How to build a sustainable infrastructure for storing the knowledge, collectively worked out in seminars, in a systematic, structured and easy to re-use way? The collaboration environments and systems successfully used so far in project work and teaching (Google Drive, GitHub, Confluence, Slack) are quite suitable for the exchange of data and information. They fall short in providing a systematic knowledge storage which can be queried flexibly, since they don’t implement the necessary knowledge graph structure.
At this point, the research question has not yet been definitively answered. The paper aims at presenting the initiated approach and at discussing the experiences so far. Therefore, the remainder of the paper is organized as follows: Section 2 provides an overview of relevant work on semantic analysis and structuring of scholarly papers content. The elaborated vocabulary to support the knowledge base is presented in Section 3, whereas Section 4 introduces the preliminary system design for the targeted knowledge base. Section 5 reflects the first implementation experiences. The paper closes with a short conclusion and an outlook on further work in Section 6. 2
There are different lines of research dealing with semantic analysis and the deployment of structured data on scholarly papers and other relevant objects of scientific environments and workflows, like conferences, proceedings, people, and projects. Table 1 gives a brief overview mentioning exemplary work in the field as well as main research objectives and findings for each of these lines. The results of meta data extraction projects like presented in Table 1 can be used as basic input for the research group knowledge base. The undertaking itself is a kind of collaborative annotation, but with a more specific focus. The increasing production of natively structured data will also support a basic input – as it looks today. However, it is not impossible that this form of publication will also support very specific, qualitative analysis questions in the future. The methods of text analysis and machine learning are the closest to the qualitative analysis of scholarly papers. Though, since a qualitative analysis is very field-specific, a high-quality training set is required. Perhaps the knowledge base presented here can serve as a training set for automatic qualitative analysis for scholarly papers in the field of Business Modeling and Knowledge Engineering from the Information Systems’ perspective. 3
Like stated above, the main objective of the required knowledge base is to support the
research group’s collective analysis of scientific publications in the field of
information systems. It is therefore obvious to structure scholarly papers according to their
main qualitative features: (i) research objectives, (ii) research methods, (iii) research
findings, (iv) future work, and (v) critical issues (comp. e.g. [
Therefore, these entities were modeled as new specific classes which nevertheless are semantically and structurally integrated in the Schema.org frame. Fig. 1 shows the high-level schema of the vocabulary. Red nodes are taken from Schema.org, the other ones are specifically modeled. Green nodes are of type Enumeration, whereas the white nodes stand for abstract concepts implementing a list structure. The vocabulary is documented on GitHub1 and referenced in LOV2.
The target system can be described as a knowledge graph, as defined in [
Knowledge Sources Qualitative judgments of
expert researchers Openly accessible structured meta data WikiData ORCID data ...
Integration & orchestration Extraction Cleaning & transformation Interlinking Validation
Knowledge Graph for Scholarly Papers
Knowledge Graph Management
Triple Store
Access LinkedData API
SPARQL endpoint Analysis & reasoning Schema Engineering & Information Knowledge
Services Research group
Wiki
Researcher
support tool
Knowledge
explorer
Information
hub
Now, the system is implemented as an out of the box OntoWiki [
The preliminary implementation as described in the previous section can be considered as a research prototype3. Since the system aims at the structured documentation and flexible reuse of seminar output of the BMaKE research group, the knowledge base is growing slowly, but continuously. At the time of writing, 35 scholarly papers from 11 publications correlated with 9 publication events (conferences) are analyzed. They are interlinked with more than 100 authors, nearly 50 organizations and places. Each month 5 to 10 new papers will be analyzed and added to the knowledge base.
The immediate support of the research group’s work allows an in-process evaluation of the support quality and a deeper elicitation of needs and requirements. The first experiences in using the system in the context of scientific seminars shows the following informal results: 1. Pure consumers of the system assessed it as very helpful in gathering deeper knowledge in the research field. 3 https://bmakewiki.th-brandenburg.de
According to preliminary assessment, a knowledge graph can be considered as a sustainable infrastructure for storing and reusing the results of qualitative analyses of scholarly papers. Even the preliminary implementation presented in this paper was evaluated as an effective (even if up to now not efficient) measure to support the work of a research group. There are three main lines of further development of the system: (i) Formal meta data which are not object of qualitative analysis must be integrated in an automatic way reusing structured data provided by open sources. (ii) A well-usable template-based form should be developed for capturing the results of the qualitative analysis. (iii) The use cases for the support of the research work must be elicited systematically and on this basis the research group wiki should be adapted. These development steps shall than be followed by a formal, structured evaluation.