Knowledge-driven systems for episodic decision support

doi:10.1016/j.knosys.2015.08.008

Knowledge-Based Systems

Volume 88, November 2015, Pages 45-56

https://doi.org/10.1016/j.knosys.2015.08.008 Get rights and content

Highlights

•
Decision support systems rely on collaborative and episodic participation.
•
Large decision knowledge bases can include varying knowledge representations.
•
We introduce an approach to implement such systems in a coherent manner.
•
The approach is based on semantic technologies and web standards.
•
A case study describes experiences with a decision support system in industrial use for years.

Abstract

The paper describes a new approach of developing and maintaining state-of-the-art decision support systems. Such systems are able to capture the collaborative work on decision problems over time. Due to the complexity of large problem spaces a multi-modal knowledge representation is proposed. For the realization of a multi-modal knowledge base we integrate semantic technologies as a fundamental layer by combining the W3C ontologies PROV-O and SKOS. The approach is demonstrated by an implementation report of an industrially deployed decision support system.

Introduction

In the past, decision support systems have been established in numerous domains. The term decision support system gathers a variety of system types helping humans to make appropriate decisions for a corresponding problem. Power [1], for instance, distinguishes the following types of decision support systems (DSS):

•
Data-driven DSS: Provide decision support based on the analysis of large amounts of data. Business intelligence systems are a typical system class of data-driven DSS.
•
Model-driven DSS: Provide decision support by using accounting, financial, representational, or optimization models. These systems provide access and manipulation of the model.
•
Document-driven DSS: Support is provided by collecting, retrieving, and (automatically) classifying large amounts of (unstructured) information. Typical system classes are document management systems and information management systems.
•
Communication-driven DSS: This system type adds capabilities to support the communication and collaboration between people working on the same task. Often this type is combined with the other types.
•
Knowledge-driven DSS: Use problem-solving capabilities to derive appropriate actions for stated problems. Expert systems and recommender systems follow the knowledge-driven approach.

In this work we focus on knowledge-driven DSS as an implementation of knowledge-based systems, where solutions (decisions) are derived from a given input of facts. The typical process of such a system is depicted in Fig. 1 in BPM notation [2], [3]: Initially a data entry activity is required to start the decision making activity, where the findings are processed by a problem solver. The derived solutions are returned in a decision output event.

Fig. 2 shows an extension of the classic decision process by allowing repeated data entry. Here, updated data is processed by the problem solver yielding a possibly updated set of decisions by incorporating non-monotonic reasoning [4].

Examples of such systems can be found in almost all domains, ranging from second opinion systems in medicine to fault diagnosis systems in the technical domain. In the past, those systems were built as monolithic applications, where single agents enter findings to derive one or more suitable decisions. Examples are found in medical consultation [5], [6], [7] and technical diagnosis [8], [9], [10].

In the context of the web and in the context of collaboration new requirements arise that motivate a rethinking of some principles of classic decision support systems.

Collaborative Use
A complex decision problem is often not solved by a single user, but it is solved by the collaborative contributions of different participants. Collaboration may have diverse faces: Users participate in the decision problem by providing or overriding important facts during a joint session. Alternatively, users solve disjoint sub-problems that in sum help to solve the overall decision problem.
Episodic Use
Complex decisions are often not taken during a single session, but the actual decision process is partitioned over time into different episodes. We align to the semantics of the term episodic introduced by Russell and Norvig [11], where subsequent episodes do not depend on what actions occurred in previous episodes. In our setting, each episode may cover a different aspect of the decision problem. Commonly, the order of the handling of the different aspects has implications of the final reasoning process.
Mixing Knowledge Representations
In traditional decision support systems a single knowledge representation is used to build the entire knowledge base. Successful knowledge representations are rules, decision trees, and Bayesian networks [12]. Complex and larger systems benefit from the use of hybrid approaches, integrating different representations into one knowledge base. Typically, a large knowledge base is partitioned into smaller knowledge spaces, where each knowledge space covers an aspect and uses a specific knowledge representation for its implementation. Here, for a single decision or fact, different knowledge representations can be continuously interweaved into a multi-modal knowledge representation.

In summary, advanced decision support systems need to deal with

•
the collaborative use of the systems by a decision community,
•
the episodic decision making of a problem,
•
and multiple knowledge representations during the decision process.

Fig. 3 depicts an updated version of the decision process. The different knowledge spaces are represented by the different sub-decision making processes. These processes can be handled in parallel by different contributors of the decision making process. A new episode is initiated by iterating the (same or different) sub-processes in the next episode. After every parallel execution of the sub-decision making processes the (final) decisions are aggregated in a subsequent process.

In the following we describe a novel approach of decision support systems by integrating different types of knowledge within reasoning for implementing the decision support. The rest of the paper is organized as follows: In Section 2 we motivate the use of a multi-modal knowledge base for building decision support systems, i.e., the knowledge formalization continuum. We also show how the use of multi-modal knowledge representations is implemented by an ontology layer. Section 3 describes the reasoning and explanation in multi-modal knowledge bases. The experiences with an industrial implementation of the presented concept is introduced in Section 4. The design decisions and experiences with building and running a collaborative decision support system in the domain of chemical safety are reported. Section 5 concludes the paper with a summary and a discussion of related work.

Section snippets

Engineering the knowledge formalization continuum

As we motivated in the introduction, complex decision support systems benefit from combining different representations instead of sticking to a single knowledge formalization. When the system needs to cover a complex domain, then it usually considers many aspects of the domain. However, for a number of practical reasons not all aspects can be included in a single knowledge base:

•
Uncertain domain knowledge: Some aspects of the domain are not well-understood in a technical sense. In practice,

Episodic decision making with continuous knowledge representations

In this section, we do not discuss the different knowledge representations and their corresponding reasoning algorithms. There exists a broad range of approaches that all have strengths and weaknesses with respect to reasoning accuracy, knowledge acquisition costs, and maintenance processes. For a thorough introduction into knowledge representation and reasoning we refer to [4], [26].

Here, we discuss the representation of the reasoning process within an episodic decision support system using

Case study: collaborative decision support for chemical safety

The previous sections introduced a general approach for the implementation of collaborative decision support systems. There are many different possibilities for the realization of an episodic decision support system using mixed knowledge representations. In this paper, we describe the implementation of a decision support system that was deployed in 2012 in its first version and since then was extended and improved continuously until now. We describe the goals, development decisions, and

Conclusions

We conclude the paper with a summary of the contributions, discuss related work, and give a brief outlook to future work.

References (54)

M. Kimura et al.
Diagnosis of febrile illnesses in returned travelers using the PC software GIDEON
Travel Med. Infect. Dis.
(2005)
J.C. da Silva et al.
Development of a knowledge-based system for cogeneration plant design: verification, validation and lessons learned
Knowl.-Based Syst.
(2014)
J. Shim et al.
Past, present, and future of decision support technology
Decis. Support Syst.
(2002)
I. Palomares et al.
MENTOR: a graphical monitoring tool of preferences evolution in large-scale group decision making
Knowl.-Based Syst.
(2014)
A.J. Williams et al.
Open PHACTS: semantic interoperability for drug discovery
Drug Discov. Today
(2012)
D.J. Power
Decision Support Systems: Concepts and Resources for Managers
(2002)
Object Management Group (OMG), 2014....
M. Dumas et al.
Fundamentals of Business Process Management
(2013)
F. Puppe
Systematic Introduction to Expert Systems
(1993)
F. Puppe et al.
Application and evaluation of a medical knowledge system in sonography (SONOCONSULT)

S. Mersmann et al.

SmartCare^tm – automated clinical guidelines in critical care

R. Milne, C. Nicol, TIGER: continuous diagnosis of gas turbines, in: ECAI’00: Proceedings of the 14th European...

R. Helfman et al.

TED-turbine engine diagnostics: an expert diagnostic system for the M1 Abrams AGT1500 turbine engine

S. Russell et al.

Artificial Intelligence: A Modern Approach

(2003)

M. Stefik

Introduction to Knowledge Systems

(1995)

W. Lidwell et al.

Universal Principles of Design

(2003)

J. Baumeister et al.

Engineering intelligent systems on the knowledge formalization continuum

Int. J. Appl. Math. Comput. Sci. (AMCS)

(2011)

N. Guarino et al.

What is an ontology?

M. Uschold et al.

Ontologies and semantics for seamless connectivity

SIGMOD Rec.

(2004)

D.E. Millard et al.

Mind the semantic gap

L. Moreau, P. Groth, Provenance: An Introduction to PROV, Synthesis Lectures on the Semantic Web: Theory and...

W3C, SKOS Simple Knowledge Organization System Reference, 2009....

W3C, SPARQL 1.1 Recommendation, 2013....

W3C, RDF – Resource Description Framework Recommendation, 2004....

P. Hitzler, M. Krötzsch, B. Parsia, P.F. Patel-Schneider, S. Rudolph (Eds.), OWL 2 Web Ontology Language: Primer, W3C...

C. Evangelou et al.

Interweaving knowledge management, argumentation and decision making in a collaborative setting: the KAD ontology model

Int. J. Knowl. Learn.

(2005)

E. Kornyshova et al.

Decision-making ontology for information system engineering

Cited by (26)

HaemoKBS: A knowledge-based system for real-time, continuous categorisation of adverse reactions in blood recipients
2021, Neurocomputing
Citation Excerpt :
There is a variety of system types, all of which have the ultimate goal of assisting humans to make appropriate decisions for practical (and often challenging) problems. Among these, expert systems and recommender systems follow the knowledge-driven approach, i.e. take into account various aspects of the domain and integrate different types of knowledge within reasoning to tackle the problem with appropriate actions [12]. Medical decision support systems were introduced as an innovative and valuable way for providing clinicians and patients with varied clinical knowledge and enhancement of the overall quality of care [13,14].
This work introduces HaemoKBS, a novel Haemovigilance decision support system for adverse reactions in blood recipients. Machine learning inference and rule-based reasoning were applied to build the underlying decision support models, namely to automatically extract evidence from different types of data included in hospital notifications and incorporate a priori expert knowledge. The ultimate aim is to dynamically learn and improve the reasoning abilities of the system and thus, be able to provide educated recommendations to hospital notifiers along with understandable explanations on the acquired knowledge. Experiments over the records of the Portuguese National Haemovigilance System from the last 10 years demonstrate the practical usefulness of HaemoKBS, which will contribute to a better depiction of the adverse reactions and to flag any incomplete notification enforcing data quality.
Ontology-based computer aid for the automation of HAZOP studies
2020, Journal of Loss Prevention in the Process Industries
Citation Excerpt :
Zhao et al. (2009) used ontologies in their HAZOP expert system. Other research groups used ontologies within decision support systems, for the support of FMEA studies, or the representation of knowledge extracted from chemical accident databases (Baumeister and Striffler, 2015; Ebrahimipour et al., 2010; Single et al., 2020). Semantic nets, frames, and ontologies can be assigned to the group of knowledge graphs.
Hazard and Operability (HAZOP) studies are conducted to identify and assess potential hazards which originate from processes, equipment, and process plants. These studies are human-centered processes that are time and labor-intensive. Also, extensive expertise and experience in the field of process safety engineering are required. There have been several attempts by different research groups to (semi-)automate HAZOP studies in the past. Within this research, a knowledge-based framework for the automatic generation of HAZOP worksheets was developed. Compared to other approaches, the focus is on representing semantic relationships between HAZOP relevant concepts under consideration of the degree of abstraction. In the course of this, expert knowledge from the process and plant safety (PPS) domain is embedded within the ontological model. Based on that, a reasoning algorithm based on semantic reasoners is developed to identify hazards and operability issues in a HAZOP similar manner. An advantage of the proposed method is that by modeling causal relationships between HAZOP concepts, automatically generated but meaningless scenarios can be avoided. The results of the enhanced causation model are high quality extended HAZOP worksheets. The developed methodology is applied within a case study that involves a hexane storage tank. The quality and quantity of the automatically generated results agree with the original worksheets. Thus the ontology-based reasoning algorithm is well-suited to identify hazardous scenarios and operability issues. Node-based analyses involving multiple process units can also be carried out by a slight adjustment of the method. The presented method can help to support HAZOP study participants and non-experts in conducting HAZOP studies.
A self-learning and self-optimizing framework for the fault diagnosis knowledge base in a workshop
2020, Robotics and Computer-Integrated Manufacturing
Citation Excerpt :
If there is no detailed mathematical model and the system has relatively few inputs, outputs, and states, the knowledge-driven approach is preferred [13]. Baumeister and Striffler designed a knowledge-driven approach to capture the collaborative work on decision problems over time [14]. Subsequently, by integrating fuzzy logic theory and data mining techniques, Xiao et al. investigated a two-stage knowledge-driven approach to generalize optimization of the energy-aware parametric for multiple machining configurations [15].
The knowledge base is an essential part of the fault diagnosis system, which is crucial to the performance of fault recognition. As the intelligence of the fault diagnosis system has made persistent advance, the increasing demands for diversity and dynamic update have posed challenges to the knowledge base. In this paper, a framework for the fault diagnosis knowledge base is proposed to address the challenges mentioned above. Firstly, a dynamic clustering model is designed using the proposed semi-supervised multi-spatial manifold clustering method to recognize attribute clusters and aggregate new types. When new types are added to this model, it is constantly updated to achieve the automatic evolution of the knowledge base for the diversity of fault. Then, a knowledge evolution model is established by the generative adversarial network algorithm to achieve self-learning and self-optimizing capabilities of the knowledge base. This method learns the distribution of knowledge elements and generates new knowledge elements to optimize the clustering model. Finally, a series of comparative experiments are carried out on bearing datasets to verify the validity of the mentioned framework and models. The comparison results indicate that the proposed method has better performance in fault diagnosis. This research can not only update the knowledge base, but also provide a feasible approach for designing an autonomous knowledge base with self-optimizing and self-learning capabilities.
A group decision making support system in logistics and supply chain management
2017, Expert Systems with Applications
Citation Excerpt :
Most model-driven DSSs are one individual user only; on the other hand, data-driven DSSs are used by multiple users across organizations. A document-driven DSS concentrates on managing, and manipulation of data in various electronic formats (Baumeister & Striffler, 2015). A collaborative knowledge base system is a communication module composing of several elements which represents various functions that the module should perform for the problem solving stages.
The paper proposes a decision support system for selecting logistics providers based on the quality function deployment (QFD) and the technique for order preference by the similarity to ideal solution (TOPSIS) for agricultural supply chain in France. The research provides a platform for group decision making to facilitate decision process and check the consistency of the outcomes.
The proposed model looks at the decision problem from two points of view considering both technical and customer perspectives. The main customer criteria are confidence in a safe and durable product, emission of pollutants and hazardous materials, social responsibility, etc. The main technical factors are financial stability, quality, delivery condition, services, etc. based on the literature review. The second stage in the adopted methodology is the combination of quality function deployment and the technique for order preference by similarity to ideal solution to effectively analyze the decision problem. In final section we structure a group decision system called GRoUp System (GRUS) which has been developed by Institut de Recherche en Informatique de Toulouse (IRIT) in the Toulouse University.
This paper designs a group decision making system to interface decision makers and customer values in order to aid agricultural partners and investors in the selection of third party logistic providers. Moreover, we have figured out a decision support system under fuzzy linguistic variables is able to assist agricultural parties in uncertain situations. This integrated and efficient decision support system enhances quality and reliability of the decision making.
The novelty of this paper is reflected by several items. The integration of group multi-criteria decision tools enables decision makers to obtain a comprehensive understanding of customer needs and technical requirements of the logistic process. In addition, this investigation is carried out under a European commission project called Risk and Uncertain Conditions for Agriculture Production Systems (RUC-APS) which models risk reduction and elimination from the agricultural supply chain. Ultimately, we have implemented the decision support tool to select the best logistic provider among France logistics and transportation companies.
A Neuro-Symbolic Approach for Anomaly Detection and Complex Fault Diagnosis Exemplified in the Automotive Domain
2023, K-CAP 2023 - Proceedings of the 12th Knowledge Capture Conference 2023
Improving decision support systems with machine learning: Identifying barriers to adoption
2023, Agronomy Journal

View all citing articles on Scopus

View full text