Actors Involved In Building A Knowledge Base

Published Date: 02 Nov 2017

This report provides a comprehensive explanation of the project undertaken in October 2012 by the author on behalf of the client, Doctus. The research focused on identifying a method by which the power of story could be harnessed to enable knowledge of the client's expert system software to be shared. Within the field of knowledge management storytelling is increasingly being seen as a legitimate and effective way to share knowledge within organisations, and it was at the request of the client that this project seek to find a way in which knowledge of the advantages of the client software, and the practicalities of its use could be conveyed to potential users. In relating this knowledge through a set of narratives to be published on the company website; it was believed that through better communication of its benefits, the software could become more highly subscribed in what is a niche marketplace.

Aims and Objectives

Following initial meetings with the client it was agreed that the goal of this project was to effectively showcase via the company website the Doctus product to potential users, while improving support for existing users. This goal was achieved by producing the following deliverables:

A set of knowledge sharing stories that communicate the practicalities and benefits of using Doctus to build a case-based knowledge base

A tool with which on completion of the project, the client could continue to facilitate the generation of user stories that share knowledge

The following project objectives were identified as the means by which the deliverables would be achieved:

Establish a narrative structure which facilitates the construction of knowledge sharing stories, which will then be used to design:

- A set of user stories to be published on the Doctus website which will enable existing knowledge of the use and benefits of Doctus to be shared and understood by users more effectively

- A set of stories that would allow website visitors to 'drill down' into further content should they want to know more in a particular area

A review of the application of visual and internet media, including social network capabilities involved in storytelling, in order to provide a recommendation as to which multimedia tools could be used to support Doctus stories in the future

Report Summary

The initial chapter of this report provides the reader with an introduction to the client company and the Doctus software, and further details the rationale for this project. The literature that was reviewed in order to build a solid understanding of storytelling within organizations is presented in chapter two; while chapter three outlines the methodological framework that was employed in creating a set of knowledge sharing narratives followed by an explanation of the data collection methods used. Chapter four will discuss the analysis undertaken after the interview stage and detail the process of narrative synthesis adopted; while chapter five will discuss the findings and present the Meta-narrative produced. Chapter six draws conclusions made as a result of having synthesized the final set of narratives, and proposes recommendations that support the client in creating user narratives in the future.

CHAPTER 1 - Scene Setting

This chapter is intended to describe the context from which this project has emerged, to provide the reader with some background surrounding the client and their product, and to explain the rationale for this project. It will begin with an introduction to Doctus in the form of a short company history, then follow with a more detailed look at the software product the client sells. The chapter closes with an insight into how this project was conceived.

1.1 The Client

1.1.1 Past

The project described throughout this report was completed on behalf of Doctus Knowledge Based System. Doctus is an expert system that was brought to life by Prof. Dr. Zoltán Baracskai during the 1970’s born of an interest in how people, particularly business executives take decisions. Doctus was designed to tackle the lack of user friendliness people experienced when using traditional 'hard coded' decision support softwares, as these typically produced highly abstract and quantitative outcomes which were often difficult for executives to understand.

Doctus applies the same principles of reasoning as these systems, but has designed a knowledge elicitation process that represents an expert's knowledge in such a way that it can be easily interpreted and understood by business people. Doctus has achieved this by developing a qualitative rather than quantitative approach that enables people to document their knowledge in their own words, so that any decision analysis outcome produced is one that they can immediately recognise and relate to. Therefore the outcome is significantly more transparent and easily understood by non-technical users.

Over the past twenty five years Doctus has been developed to execute several types of reasoning and deliver increased functionality, and has been used successfully by a number of large clients for a variety of decision scenarios. National oil and utilities companies, broadcasting firms, banks, hospitals, manufacturers, local government bodies and educational and research institutes have all used Doctus for a wide array of decisions. From wanting to know where best to build a nuclear power plant, establishing how one might identify a high risk customer, to deciding where to locate the pins needed to repair a radial distal fracture; the knowledge of experts within a range of specialist areas is consistently in demand.

1.1.2 Present

At present Doctus is used by a number of universities and research institutions throughout Europe but as the success of its implementation is heavily reliant on the Doctus consultation team, which currently consists of solely Dr. Dörfler; the software has not been adopted as heavily in industry as the team would like in order to generate the revenues to improve the software’s online presence, and to update the current graphical user interface to reflect the standards expected by today's users.

1.1.3 Future

It is the client’s intention that given the right degree of investment, DoctuS software could be available to download for free where support from consultants would be provided for a fee. This would enable a versioning strategy to be developed where three levels of service delivery could be offered: Formal (face to face)consultancy, Online consultancy (via Skype and email) and also a 'self-taught' version where users could be supported by online tutorials, videos, and have access to help during crucial project initiation and fine tuning stages. In order to support this strategy the website would be improved to allow continuous administration to be achieved through the introduction of features such as a regularly updated blog, customer review facility, and through more interactive video content and social network site integration.

1.2 Doctus

1.2.1 What is DoctuS ?

Doctus software is a knowledge based expert system shell that models the knowledge of an expert within a particular field. This knowledge is used to either evaluate the alternative outcomes of a decision or to identify the rules applied in making decisions, to support the choice of a decision taker. DoctuS is a Latin work meaning 'learned', 'wise', expert' and to 'teach' (www.latin-dictionary.org) which reflects the purpose of this software; to help people to discover more about what knowledge they possess. Turban et al. (2007: p756) describe an expert system as:

"A computer system that applies reasoning methodologies to knowledge in a specific domain to render advice or recommendations, much like a human expert. An ES is a computer system that achieves a high level of performance in task areas that for human beings, require years of special education and training."

As DoctuS utilises knowledge rather than data or information it is an example of a Decision Support System (DSS) and could also be considered a Strategic Information System (see Figure ) as it can be used to support executive strategy.

A knowledge based expert system is one that uses Artificial Intelligence (AI) to solve problems by utilising a repository of expert knowledge from which knowledge can be retrieved in response to particular queries, and through justification and learning allows knowledge to transfer between parties (Akerkar & Sajja, 2010). Specifically, knowledge based systems aim to aid decision making, learning and action through problem solving (Akerkar & Sajja, 2010). The goal of AI is to create a computer that 'thinks' like a human in a specific field, or one that resembles the type of behaviour that we would consider intelligent if it was carried out by a human (Dörfler, 2012) .

Progression of Business Information Systems

There are two main approaches to AI; strong or connectivist AI, and weak or symbolic AI. Symbolic AI is an approach which aims to apply rule-based systems that express knowledge as logical statements in order to act as a problem solving tool. By achieving a result for a specific task that would be similar to that when intelligence is applied by a human carrying out the same task, this approach seeks to employ AI without the need for a full understanding and recreation of human thinking and cognition; which is the goal of many systems adopting the Connectivist approach (Dörfler, 2012). DoctuS is an example of weak or Symbolic AI in action.

There are three main roles which must be adopted when building a Doctus knowledge base to establish the rules applied when making a decision; the knowledge engineer, the domain expert and the decision taker (see figure). The expert will have significant and specialist experience within a certain field, perhaps acting as a structural engineer or cardiologist for example. The decision taker could be a business leader who wishes to understand this expertise to better support an important decision. Those who use DoctuS are most often executives who are making a substantial financial investment in implementing a decision, or those who seek to achieve significant financial benefit from a decision alternative being successfully applied.

Actors involved in building a knowledge base

As we cannot transfer knowledge from an expert straight into a system one must gather the knowledge from the expert using a technique called 'knowledge acquisition'. The knowledge engineer will work with the expert to have them describe their decision situation using Attributes and Values which can be used by the engineer to populate the system shell thus representing the expert’s knowledge. This process involves the engineer teasing out a blend of tacit and explicit knowledge from the expert over time.

Attributes are nominal scale variables so have no ranking order. We use this term rather than 'variables' which are quantitative. Attributes are used to describe each aspect of the decision which must be taken into account, for instance when buying a car we may consider attributes such as colour, speed and fuel economy.

Values are expressed on an Ordinal scale so are used to rank an attribute but not necessarily with equal intervals separating them. The expert identifies values for each attribute by articulating the most and least favourable extremes often with an intermittent degree of variation. As we do not consider decision alternatives having attributed them a weighted numerical value, we are in a non-quantitative school of thinking. For instance we do not rate how much we like something by saying "I like this 1.5%", we say "I don't like it a great deal" or that it is bad or good.

Doctus is capable of handling three types or reasoning: Rule-Based, Case-Based and Extracted Rule-Based reasoning. As this project focuses on the application of Case-Based reasoning, this approach will be further explained (please see Appendix for a description of both Rule-Based and Extracted Rule-Based reasoning).

1.2.2 Case Based Reasoning

According to Taylor (1990: p131) knowledge consists of: 'knowledge mediated by deduction from known principles' and also 'the knowledge of the principles themselves, acquired by induction'. Therefore where an expert cannot articulate the ‘principles’ or rules which they apply in order to make a decision, yet can describe a long history of decision cases from experience; these cases can be used to identify the rules applied using inductive or Case Based Reasoning. Inductive reasoning within DoctuS is applied when experts describe different cases (instances of decision making) and explain the outcomes of each of these cases (the results of the decision which was taken). For instance a Doctor may explain several examples of where they have made a certain diagnosis.

Algorithms within DoctuS infer the rules applied to each case, to formulate a rule set that suggests the attributes which have been most informative in helping the expert reach their decisions in the past. For example the Doctor may discover that a patient's age or gender may be a significant contributing factor in diagnosing a certain problem. This type of reasoning requires the expert to be able to recall multiple cases of decisions being made and so demands extensive experience on the part of the expert. Because it involves the expert exercising to a large degree the tacit knowledge they have acquired from experience over time, this would constitute an example of a Routine or programmed decision being made (see figure 3).

Actors involved in building a knowledge base

1.2.3 The Benefits of DoctuS

As DoctuS is a software shell it avoids the inflexibility of systems where the knowledge of an expert is coded directly into the software, meaning that DoctuS can be used to model the knowledge bases of multiple people, where each set of cases is personalised and the qualitative outcomes can be more easily understood and verified by the expert. As the software uses symbolism to explain its logic in the form of an illustrated graph, this also allows for easier interpretation of the reasoning by non-technical users. As the DoctuS user interface was designed to reflect the look and feel of Microsoft Excel it is very user friendly given that most computer users have experience of using this spread sheet software. Although as already mentioned this interface needs updating to reflect current user standards (please see appendix for full explanation of the shortcomings of DoctuS).

Doctus provides users with added functionality in that decision cases can continue to be as added to a knowledge base, via an HTML page hosted on a corporate intranet or portal for instance, to reflect the growing experience of the expert and the changes within their decision making environment.

This prevents the cases covered in the base from becoming obsolete as new instances are being added which helps to maintain ‘fresh’ and timely results. Knowledge in the form of data can also be imported from external sources and completed knowledge bases can also be exported for integration with a client’s own system, meaning that it can act as an on-going knowledge portal from which a client can support future decisions.

Perhaps the most salient benefit to users is that an expert may realise their knowledge as a result of the process where they may learn of the attributes which are most informative in helping them reach their decision; as in the example of the Doctor already mentioned. As the knowledge engineer encourages the expert to express their experiences verbally, some tacit knowledge becomes explicit during the process. Therefore DoctuS helps people to discover knowledge they perhaps didn’t know they were applying. This experience of knowledge realisation and discovery supports the expert in making future decisions in a more self-aware and enlightened state, as they can live their life with a greater understanding of the aspects which have influenced their behaviour in the past. This transparency allows decision makers to be comfortable in taking decisions going forward, and may allow an expert to explain to others the rationale behind their decision making.

1.3 Project Rationale

It was felt by the client that given the relative perceived complexity of expert-systems, there exists potential for the benefits and practical application of Doctus to be more effectively showcased. At the moment there is to some degree a lack of clients adopting regular use of the software, and it was felt by the Doctus consulting team that online support for the software could be improved to enable the package to become more highly subscribed in what is a niche marketplace.

The Doctus consulting team were therefore looking to update their website so that the benefits and practical use of the software could be more easily communicated to existing and potential users, as it was believed that this may improve the appeal of the product. As the client was aware from his understanding of current knowledge management literature that experiences can be powerfully conveyed when presented in narrative form, I was approached to establish a way in which Storytelling could be used as a vehicle with which to share existing knowledge of how Doctus is used in practice.

1.4 Chapter Summary

This section of the report has outlined the history of Doctus as a company and as a software, and has provided an overview of the purpose of Doctus and the way in which it can be applied to the business context. It has explained that Doctus is an expert knowledge-based system which is underpinned by a form of Artificial Intelligence, and that its' aim is to support people in business environments with a variety of decision making scenarios. I have introduced that rationale for this project as intending to support greater adoption of the software through better understanding of its use and benefits through the communication of knowledge sharing stories. The following chapter will present the literature that was reviewed to support the project in sharing knowledge through storytelling.

Appendix 1: Doctus Reasoning

Rule Based Reasoning

Deductive or Rule Based Reasoning is where knowledge is "mediated by deduction from known principles" (Taylor, 1990: p131). Within DoctuS this is achieved where the explicit rules of a decision are articulated by the expert then added to the shell. For instance the expert will connect the values and attributes by applying ‘if’ and 'then' rules, for instance when buying a car; "IF the car is fast AND it is affordable AND it is safe, THEN I will buy it". This helps the expert to clarify their expectations explicitly, and articulate the rules between their expectations so that they can apply deductive reasoning. These rules can be used to connect the values and attributes which are loaded into the software shell, once applied these rules will generate an evaluation of the possible decision outcomes.

This type of reasoning is used when the expert does not have sufficient experience to produce a high enough number of cases (or instances of decisions being made) within their field, this may also be referred to as an original or non-programmed decision, as it is one that the decision taker has never had to make before (see figure 3).

If the knowledge engineer has successfully captured the knowledge of the expert then the system will apply the same evaluation of cases that the expert would, meaning that when taking the original decision one can propose a particular decision with confidence. When the decision outcome suggests a number of alternatives, we could either define additional rules which is called 'fine tuning' or we could just take the first one that satisfies our expectations – thus employing Simon's theory of 'bounded rationality' (Simon, 1997) where the most rational choice is selected given the limitations of the decision scenario as understood by the expert.

Extracted Rule Base Reasoning

Reductive reasoning is applied when DoctuS generates "rules that describe the cases of experience" (www.doctuskbs.com). This method of reasoning extracts the rules which were used by an expert as demonstrated throughout an existing case based knowledge set. These can then be used to create an instance of rule based reasoning which applies only those attributes that were the most informative in the outcomes of the expert’s past cases, where we are reducing the number or attributes employed as those that have no bearing on the outcome can be shed. Therefore we "are learning from the cases of experience" (www.doctuskbs.com), by combining both approaches and reducing the number of attributes needed to arrive at a decision outcome.

Appendix 2: The shortcomings of DoctuS

DoctuS requires a large amount of time to acquire attributes and cases from experts, and some may find it hard to articulate the nuances in their decision making processes. It is also recommended that the decision taker be involved in the part of the knowledge acquisition stage in order for them to better accept the process and the outcome. This time very quickly translates into cost. A Doctus client would be expected to pay upwards of £20k to cover the consultant and expert hours required to support the elicitation process. It should be noted that it would be unlikely that a knowledge base would be programmed correctly and perform successfully without the input of the DoctuS consultation team. Smaller companies may also find it hard to identify a member of their staff who has the experience in one area to be called a domain expert as often their people perform a number of roles so lack the sufficient experience in one specialism to be classed as having an 'expert' level of knowledge.

Another issue which would need to be addressed when employing case based reasoning is that the knowledge engineer employed to elicit the expert’s knowledge would ideally be able to understand the common language of the expert’s domain. The terminology used by those who operate in the world of finance, law, engineering or even catering and carpentry for example are unique to those who perform in each domain, and being able to recognise and interpret this language makes eliciting tacit knowledge easier for the engineer and less time-consuming for the client. However having available an engineer suitable in this way is not often possible.

Appendix 3: Project Proposal

'Showcasing Doctus:

How to explain the use and illustrate the benefits of Doctus to existing and potential users'

Project Overview and Background information

Doctus is an expert system shell that can be populated with representations of the knowledge of individuals, in order to assist those taking decisions within an organisation. Having met with Dr. Viktor Dörfler who is also the project supervisor, it is felt by the Doctus consulting team that online support for the software could be improved to enable the package to become more highly subscribed in what is a niche marketplace. On meeting with the client it was asserted that the website and its content needs to be refreshed in order to breathe new life into its success as a product, and to enable the consulting team to grow

The Doctus development team is looking to update the website and associated user documents (including multimedia) to allow the benefits and practical use of the software to be more easily communicated to existing and potential users. The client has asked myself to design a set of stories which will allow the way in which Doctus can be used to be more easily understood by the users, and to produce an accompanying set which will use Doctus case studies to communicate the benefits of the software

At the initial client meeting it was agreed that I would consider the online nature of the project in my designing the content to be published, and seek to leverage available multimedia tools to increase the positive impact of the stories. It was also discussed that should there be features which would allow more customised versions of the software to be marketed to different segments in the market, for example executives and systems analysts, there should be recommendations made as to how this could be delivered.

As the project seeks to investigate and discuss the user response to changes made to the website, the research could provide academic value through an improved understanding of expert systems through online multimedia storytelling. As organisations move more of their resources online and as our personal world too becomes more virtual, it seems pertinent to investigate how this experience can be improved to foster knowledge sharing.

Project Aims and Objectives

The goal of this project is to refresh the Doctus website in order to more effectively showcase the product to potential users, while improving support for existing/new users. This will be achieved through meeting the project aims:

Produce a set of knowledge sharing stories that communicate the practical use and benefits of Doctus

Produce a tool which on completion of the project, the client could continue to facilitate the generation of user stories that share knowledge

The project deliverables would subsequently be achieved by meeting the following objectives:

Establish a narrative structure which facilitates the construction of knowledge sharing stories, which will then be used to design:

- A set of user stories to be published on the Doctus website which will enable existing knowledge of the use and benefits of Doctus to be shared and understood by users more effectively

- A set of user stories that would enable a potential user to 'drill down' into further content should they want to know more in a particular area

To review the application of visual and internet media including social network capabilities involved in storytelling, in order to provide a recommendation as to which multimedia tools could be used to support Doctus stories in the future

Personal Learning Objectives

In delivering this project there are a number of personal objectives which I hope to achieve by fully exploiting the opportunities that the dissertation process will present me with:

I see the project as a ‘learning’ as well as a 'doing' experience. I intend to learn about how to manage a real life project first hand, and thus improve my organisational and planning skills. Although I have had input into group projects and in meeting employer and university deadlines, I see the delivery of this project as a challenge due to the level of unpredictability involved and the weight that its completion will have on my future career path. In order to gradually improve my ability to manage the project, I intend to take time out for regular refection which will enable me to act upon the situations I find myself in keeping the project on track.

I hope to maintain a realistic and effective relationship with my client. As I have the opportunity to progress in to a consulting role with an accountancy firm on graduation, I am hoping to maximise my understanding of how to manage such a relationship while developing my communication and listening skills. As one of the representatives of the client is also my supervisor I intend to be mindful of when I approach him in a client/employee capacity, and when asking for feedback on my performance in both student and employee roles which I intend to use to adapt my approach accordingly.

I intend to improve my academic understanding of both expert systems and storytelling throughout the lifetime of the project.

Relevant past studies

In terms of general approaches to consulting as well as providing examples of management science tools I feel that both 'Foundations of Business Analysis' and 'Consulting and Simulation' classes will provide useful in forming an approach to structuring the problem, and in my understanding the relevance of spreadsheet modelling and Bayes theorem in the context of Doctus.

While studying for the third year class 'Knowledge Management' I was introduced to the importance of human knowledge in organisations and the development of personal knowledge, which I feel is relevant given that one of the objectives of this project is to enable the sharing of knowledge and the increase of user knowledge. As I became aware of the advantages of the 'learning organisation' and web 2.0 tools in enabling communities to evolve, I feel that I have the background understanding to make meaningful recommendations to the client given the online context. As I have an interest in the value of personal knowledge being recognised by organisations, developed from personal experience and my studying Knowledge Management, I feel that this is a unique opportunity for me to build on my understanding of expert knowledge and how it can be applied.

Although this project is the first experience I have had with Doctus as I was unable to take part in the class 'Working in today's virtual world' due to being abroad on exchange, I have access to the relevant class materials and will use those covering Doctus to increase my elementary understanding of the software. I intend also to draw from the introduction to expert systems which was part of 'Information Systems in the Knowledge economy'.

I took part in an E-Commerce class while studying in Hong Kong. This has given me an insight into basic web design and how to measure website usability which I imagine will stand me in good stead for understanding web design literature and making recommendations as to the features to be included in Doctus site.

My current studies in Project Management have provided me with the an introduction to the tools available to assist the planning of my project, while helping me to stay mindful of the need to plan and control the project whilst considering time and resource constraints.

Methodology

In carrying out my project I intend to use the management science methodological cycle - a framework which was introduced to me in second year as part of the Consulting and Simulation management science core class. I feel that this is an appropriate model to use as it will provide me with a sound structure to refer to as I am progressing throughout my dissertation, and as it emphasises the importance of client interaction in the consultation environment, it will be useful in keeping my focus on delivering for the client. In terms of collecting data and structuring the problem from user and client perspectives, I intend to observe the use of Doctus by expert (client representative) and 'novice' student users. In order to supplement the understanding this will provide me with I plan to attend Doctus specific lectures, and interview users regarding their understanding of Doctus. To be able to apply the principles of storytelling which benefit prospective users to the Doctus context, I will need to have a sound understanding of personal knowledge and storytelling and how it can be applied in practice. I intend to develop this knowledge by reading literature produced by authors such as Michael Polanyi, Herbert Simon, Stephen Denning and David Boje.

Work Plan - See attached Gantt Chart

Key tasks:

Produce a literature review: to have a reliable grasp of knowledge surrounding knowledge based systems (particularly Doctus) and storytelling, upon which I can recommend a method for narrative creation, and suggest how this method can be supported by the client in the future.

Design a story structure: with which to create stories

User interviews: to collect data on understanding of Doctus use and to act as material for story creation

Create a set of knowledge sharing stories: use the findings from student interviews to create a set of stories that share knowledge about how Doctus is used in practice.

Contingencies:

In planning my time until submission I have considered time that must, or will inevitably be taken out from working on this project. For instance I have allocated a weeks' break for Christmas and New Year and also two weeks to dedicate solely to the January exam diet. I have also considered that the Easter break precedes the final submission deadline, so have planned to have a full draft submitted to my supervisor a week prior to the holiday commencing, allowing him time to review the work and propose corrections. Unfortunately due to an unavoidable late start to my work, given the estimated task durations I have considered, I will be unable to find contingency time should I fall ill or any unforeseen setback occur.

Outline of chapters:

Introduction

Background Information - an introduction to expert systems and Doctus

Literature Review

- Personal Knowledge

- Sharing Knowledge

- Stories and Storytelling

Methodology

- Project Framework

- Data Collection

Analysis & Story Synthesis

Findings

Conclusion & Recommendations

Personal Reflection

Practical & Ethical Issues

As I intend to interview university students on their opinion of existing Doctus materials, as well as any changes made to the Doctus user guide and website, I will submit my proposal for consideration by the Ethics Committee.