The Use of Semantic Web Technologies for Decision Support - A Survey

Paper Title: 
The Use of Semantic Web Technologies for Decision Support - A Survey
Eva Blomqvist
The Semantic Web shares many goals with Decision Support Systems (DSS), e.g., being able to precisely interpret information, in order to deliver relevant, reliable and accurate information to a user when and where it is needed. DSS have in addition more specific goals, since the information need is targeted towards making a particular decision, e.g., making a plan or reacting to a certain situation. When surveying DSS literature, we discover applications ranging from Business Intelligence, via general purpose social networking and collaboration support, Information Retrieval and Knowledge Management, to situation awareness, emergency management, and simulation systems. The unifying element is primarily the purpose of the systems, and their focus on information management and provision, rather than the specific technologies they employ to reach these goals. Semantic Web technologies have been used in DSS during the past decade to solve a number of different tasks, such as information integration and sharing, web service annotation and discovery, and knowledge representation and reasoning. In this survey article, we present the results of a structured literature survey of Semantic Web technologies in DSS, together with the results of interviews with DSS researchers and developers both in industry and research organizations outside the university. The literature survey has been conducted using a structured method, where papers are selected from the publisher databases of some of the most prominent conferences and journals in both fields (Semantic Web and DSS), based on sets of relevant keywords representing the intersection of the two fields. Our main contribution is to analyze the landscape of semantic technologies in DSS, and provide an overview of current research as well as open research areas, trends and new directions. An added value is the conclusions drawn from interviews with DSS practitioners, which give an additional perspective on the potential of Semantic Web technologies in this field; including scenarios for DSS, and requirements for Semantic Web technologies that may attempt to support those scenarios.
Full PDF Version: 
Submission type: 
Survey Article
Responsible editor: 
Krzysztof Janowicz

Review 1 by Naicong Li:

The new version and the author's response letter look fine to me.

Perhaps just one more change:
p. 18, " well as efforts for providing reusalble online ontologies for this field [62]"
change to: "... as well as efforts in using ontologies to organize and better access spatial decsion support resources (data, models, tools) [62]"

Paper and response letter have been accdepted by the editor. Some incomplete references have to be corrected prior to publication.

The reviews below are for the initial submission, while the PDF file contains the resubmitted manuscript.

Review 1 by Naicong Li:

This paper presents the results of a literature survey of Semantic Web technologies used in decision support systems (DSS), and the results of interviews with DSS researchers and developers. It identifies the areas where semantic technologies are being applied with respect to various aspects of DSS functionalities, and presents an assessment of the current state of this intersection of the two fields. The paper is well structured, with a clear description of the methods used in the literature survey and interviews, a clear analysis of the data collected. Its discussion on the state of the art is informative, and the paper is making a good contributions to the relevant fields.

In my view, the main areas where the author could improve on this paper are as follows:

The author did a good job covering a number of DSS application areas such as business intelligence, healthcar, emergency management. However the author might want to extend the coverage to planning and decision making in natural resource management, and urban and regional planning, with their numerous sub domains, for example, water resource management, forest management, endangered species recovery planning, transportation planning, land use planning, to name just a few. Decision support plays an extremely important role in these areas, and the DSS in these areas are often referred to as spatial decision support systems (SDSS), planning support systems (PSS), etc., which are a subset of DSS and have an explicit spatial dimension (of course, many DSS included in the current papers survey have a spatial dimension as well). Below are links to a few papers (although not from the list of journals that the author looked into) that are relevant for DSS and Semantic Web:, Villa et al., Modelling with knowledge: A review of emerging semantic approaches to environmental modelling
• Ceccaroni, L., Corte´ s, U., Sa`nchez-Marre` , M., 2004. OntoWEDSS: augmenting environmental decision-support systems with ontologies. Environmental Modelling & Software 19 (9), 785–797. , Jelokhani-Niaraki and Malczewski, A User-centered Multicriteria Spatial Decision Analysis Model for Participatory Decision Making: An Ontology-based Approach , Boerboom, Integrating Spatial Planning and Decision Support System Infrastructure and Spatial Data Infrastructure (where you can find many other related papers in the reference section) Jung & Sun, Ontology-driven Problem Solving Framework for Spatial Decision Support Systems, Zhang et al., The framework of a geospatial semantic web-based spatial decision support system for Digital Earth.

A related topic the author may want to look into is the decision making process workflows and methods typically used in structured or semi-structured large-scale planning processes, and investigate whether there are research efforts going on in applying semantic web methods and technologies to facilitate such workflows and methods. The paper by Jelokhani-Niaraki and Malczewski above touches on this subject.

To collect literature in the areas mentioned above, the author should perhaps include "planning support" in the keyword list used in literature collection (section 2.1.1). The list of journals could be expanded also. If the number of additional relevant articles turns out to be not substantial, it would still be a fact good to be pointed out in the paper.

Editorial comments:

Section 2.2: It would be nice to make it more obvious in the graphs in this session that the 2012 numbers are only for a partial year. This fact is stated in the text but not reflected in the graphs.

Section 2.2: The legend part of Figure 5 and 6 is too blurry to read.

Section 3.2.2: This section is about information filtering and selection, but there is a paragraph on information aggregation, which seems to belong to the section 3.2.3.

Section 3.2.6: If this section is meant to cover the last two items in the "list of feature categories" in Section 3.1.2, then the section title needs to reflect this.

Section 4: Citation [58] is not a good example for LOD and spatial data. There are other much better references for this. Citation [58] would fit better in Section 2.2 in the paragraph where semantic annotation of models is discussed.

There are also a number of sentences that need to be edited for grammatical errors (e.g. Section 3: (before section 3.1): "The main aim of the interviews being to survey what challenges are perceived by DSS researchers and practitioners today, and …" should (at least) have the following modifications: "The main aim of the interviews is to survey what challenges are being perceived by DSS researchers and practitioners today, and …". I suggest a careful edit of the entire paper.

Review 2 by Olivier Bodenreider:

This paper presents a review of the use of semantic web technologies for decision support, based on a review of the literature in the past decade or so (60 articles) and the interview of 7 specialists from Europe and the US. The main conclusion seems to be that the penetration of SW technologies in DSS remains modest at best, in spite of their relevance to several aspects of DSS.

This reviewer is not a specialist of DSS, therefore only general comments and suggestions will be provided here.
This review article is generally interesting, well written and seems to provide a comprehensive analysis of the use of SW technologies for DSS. In addition, a gentle introduction to DSS is provided. The paper is relatively long but easy to read. The list of references is extensive. The review methodology is classical and appropriate in this context. I would recommend this manuscript for publication with no need for any major revisions.

Minor comments
The title could make it even clearer that this paper is a review (e.g., though the use of a subtitle).
This paper fails to discuss probabilistic models and the difficulty to mix up symbolic knowledge (e.g., description logics-based ontologies) and statistical knowledge as a potential reason for the limited adoption of SW technologies in DSS.
Intellectual property restrictions associated with some knowledge sources could be discussed as well. This is the case in particular in the clinical domain and this is the reason why few knowledge sources are publicly available in ontologies or as SPARQL endpoints. The lack of native access control in SW technologies contributes to their lack of adoption in commercial DSS application.
Finally, SW technologies are presented as the ultimate solution for data integration. In fact, Linked Data mostly supports the linkage of datasets in which URIs are shared. However, given the lack of standardization in minting URIs and the variety of identifiers available in domains such as the life sciences, a substantial effort is still necessary for the meaningful integration of heterogeneous datasets.

Review 3 by anonymous reviewer:
The paper by Eva Blomqvist fills the gap by providing well-structured and clearly written review of published research in the areas of Decision Support Systems (DSS) and Sematic Web. The focus of the review is on cross-fertilization between DSS and Semantic Web. The author has structured her review by first conducting a keyword based search of published work in (1) online indexing services, (2) publisher databases, and (3) individual journals/publication series, and then expanding upon the literature review by a series of open-ended interviews with researchers and DSS/Semantic Web developers outside academia. The results of the review are followed by the discussion of the current state-of-the-art and needs in developing Semantic Web to become instrumental for enhancing DSS.

Some clarifications are needed before the paper can be accepted for publication. First, in the description of the literature survey methodology the author uses plural for online indexing services and publisher databases while only one indexing service (Google Scholar) and one publisher database (SpringerLink) were used. The question that begs the answer is: why only Google Scholar? Why not other comprehensive indexing services such as IEEE Explore and Web of Science (Web of Knowledge)? Then, in light of the singular indexing service used for the review a follow-up question arises: how comprehensive the presented literature review is?

I disagree with the use of "statistical approaches" in the sentence on page 15: "General statistical approaches, such as free text search with Google-like methods, might not work as well as in the general case since keywords are highly specialized and alone may not be enough to discriminate between documents." In my understanding "free text search" is not an example of statistical approach.

Submission in response to



I enjoyed reading this article, even more so as I agree that many types of DSS will profit by relying on Semantic Web Technologies.

One comment on the Interview Study.

Conclusions regarding the needs of DSS practitioners are based on the interviews of seven partners from the Swedish defence, training and health sector. It would enforce the generality of the conclusions if related work was mentioned.

Such work has been done within EU-funded research projects.

The VALUE-IT project [1] was dedicated specifically to analyse take-up of Semantic Technologies for the Enterprise. They carried out 50 in-depth interviews of up to 625 IT decision makers and generated a significant number of reports and figures. Unfortunately, their website is currently offline.

VALUE-IT findings regarding the use of semantic technologies in relation to BRMS and DSS are also mentioned in ONTORULE Deliverable D8.4 "Market Intelligence Report" [2]. In [2], the implementation gap analysis (VALUE-IT D3.2, 2010) together with IDC reports 2008-2010 are taken as starting point to claim that Business Rule Management Systems and Semantic Technologies would profit by merging in a new type of Decision Support Systems.

Another survey containing relevant info is that of Frithjof Dau [3], with some details on the use of semantics in Business Objects.

The above-mentioned surveys confirm the observations from the interviews and complement them with views from other applications areas and sectors (BRMS, manufacturing).