The process of creating the PhD thesis:
Representation of Industrial Knowledge
- as a Basis for Developing and Maintaining Product Configurators
By Anders Haug, in 2007,
at Department of Manufacturing Engineering & Management,
Technical University of Denmark
The process of creating the PhD thesis:
Representation of Industrial Knowledge
- as a Basis for Developing and Maintaining Product Configurators
By Anders Haug, in 2007,
at Department of Manufacturing Engineering & Management,
Technical University of Denmark
Agenda 1) Why write a paper-based PhD
2) The work process
3) My thesis
4) Methodological reflections
5) The biggest crisis and the best experiences
6) Questions
1) Why I chose to make a paper-based PhD
Writing a paper-based PhD thesis Against writing a paper-based PhD:
Need for having a clear focus very early – the first papers cannot be changed
Difficult to conduct large studies
Several small conclusions, i.e. a more unclear contribution
Much harder to create consistency
More time-consuming
Greater chance of not completing the project
For writing a paper-based PhD:
At the beginning of my PhD thesis I was 100% sure that I wanted to have an academic career.
The more papers, the better for my career as a researcher.
May be easier to divide into smaller pieces.
2) The work process
Work process The process
0-1: Produce papers (within the topic)
1-2: Create papers that ensure consistency
2-3: Write the theses and complete the last papers
Key points
- Get a clear idea of the focus of the thesis as early as possible
- Be aware of methodology from the beginning; if done ok, this can be consistent with different lines of theory of science
- Do not wait to long before thinking about the thesis as a whole
3) My thesis
Basis "...the area of knowledge acquisition in the context of product modelling and product configuration is quite untouched, although generic methods exist for knowledge acquisition... Hence, there is a need for new and more specific methods for how the knowledge of complex products is formalized and modelled into a product model that can serve as a basis for a product configuration system." (Hvam et al., 2006)
Focus Purpose of the thesis:
"To improve the methods, techniques, tools and understandings that form the basis for representation of configuration knowledge"
Structure
Research Questions 1) Does the use of the term 'tacit knowledge' in configuration literature comply with the original meaning of the term, and does it make sense to apply this term in configuration research?
2) Can the knowledge/information that a domain expert possesses and delivers to a knowledge engineer in a configuration project be categorized in a better way than the tacit-explicit knowledge distinction?
3) What are the limitations of applying the PVM formalism, and how can the formalism be altered in order to solve such limitations?
4) What are the actual differences between using PVMs and class diagrams for modelling problems in configuration projects?
5) How can the migration of information from PVMs to class diagrams be avoided while not losing the benefits of the application of both techniques?
6) How can models with overlapping information in configuration projects be maintained, while avoiding the necessity to update the same information in several places?
7) What are the necessary definitions for the creation of a documentation system that supports the CPM-procedure?
Relations between research questions/papers (a process)
The included papers A. Haug: 70 %
A. Degn: 20%
B. Poulsen: 10%
L. Hvam: Approval of manuscript Haug, A., Degn, A., Poulsen, B., and Hvam, L. (2007): "Creating a documentation system to support the development and maintenance of product configuration systems", in Proceedings of the 2007 WSEAS International Conference on Computer Engineering and Applications, Queensland, Australia, Jan. 17-19, 2007. C3 A. Haug: 100 %
L. Hvam: Approval of manuscript Haug, A. and Hvam, L. (2006): "CRC-cards for the development and maintenance of product configuration systems", in Customer Interaction and Customer Integration (Proceedings of the Joint Conference IMCM'06 & PETO'06, Hamburg, Germany, June 22-23, 2006), GITO-Verlag, Berlin. C2 A. Haug: 100 %
L. Hvam: Approval of manuscript Haug, A. and Hvam, L. (2007): "The modelling techniques of a documentation system that supports the development and maintenance of product configuration systems", International Journal of Mass Customisation, Issue 1/2, Vol. 2. C1 A. Haug: 100 %
L. Hvam: Approval of manuscript Haug, A. and Hvam, L. (2006): "Merging models with different perspectives on product configuration knowledge", in Research in Interactive Design, Volume 2 (Proceedings of Virtual Concept, Cancun, Mexico, Nov. 26 - Dec. 1, 2006), Springer-Verlag, France. B4 A. Haug: 100 %
L. Hvam: Approval of manuscript Haug, A. and Hvam, L. (2007): "Product Structured Class Diagrams to support the development of Product Configuration Systems", in Proceed...
Summery of contributions 1-6 1) An overview of relevant research carried out at four of the most important groups involved in configuration research.
2) A clarification of the meaning of some important terms and concepts that are commonly applied in configuration research ("configuration", "configurator", "ETO vs. mass customization").
3) A possible solution to the conflict between scientific ideals (critical rationalism) and the way configuration research is actually carried out.
4) Clarification of the meaning of the concept of 'tacit knowledge' and analysis of the unfortunate consequences of the current use in configuration literature.
5) A classification of the information possessed and delivered by a domain expert in a knowledge acquisition situation.
6) A notation technique, named PFDs, which has shown to be useful in special cases.
Summery of contributions 6-12 7) Insights into the differences between using PVMs and class diagrams to create models in configuration projects.
8) A layout principle for class diagrams, named PSCDs + investigations that indicate that the major advantages of both diagrams have been achieved.
9) A modelling principle that allows separate models with overlapping information to be maintained in a common model + definitions and test of a software solution.
10) A definition of a documentation system that supports the modelling techniques of the CPM-procedure.
11) A generic CRC-card definition that allows companies to adapt the extensive basic layout to individual needs.
12) The first software prototype that supports the modelling techniques of the CPM-procedure in an integrated fashion.
4) Methodological reflections
Scientific ideal of the research community (DTU) ● Quantitative and qualitative studies
● Often the application of action research
● Not stating hypotheses that are attempted falsified
● Continuous development, rather that "Popperian" experiments
● Study how humans apply models, methods and techniques
● The use of human factors as explanations
● Etc. Critical rationalism ?
Logical positivism Problem 1 (logical): Stating under which exact conditions a generalization is a sound inductive inference. This is because inductive arguments refer to prior knowledge, which also has to be derived by induction and based on new conditions, which also need to be justified, and so on.
Problem 2: Inexact measurements. Although this is accepted by natural scientists, it raises the problem of how exact laws can be derived from inexact evidence.
Problem 3: 'The induction problem'. The induction problem refers to the fact that since general scientific laws go beyond a finite number of observations, these can never be proven.
Problem 4: Defining criteria for when a hypothesis is valid.
"Failed" attempts to save positivism: Recognizing that it may not be possible to establish absolute truths about reality, later positivists have resorted to the use of probabilities. Since, a finite set of observations divided by an infinite number of not yet conducted observations implies a probability going towards zero, more refined ways of dealing with probability have been created. However, none of these represent unproblematic solutions, e.g. 'Bayesian approaches', where some are criticized for implying subjective probabilities.
Critical rationalism (Popper) Popper defines an acceptable 'empirical theoretical system' as fulfilling:
It consists of a set of formalized statements (if they are not formalized, they cannot be used for deductions)
The set should be mutually consistent (i.e. no contradicting statements)
Some of the statements must be synthetic (the truth of the predicate may not be a consequence of the nature of the subject, e.g. in an analytical statement such as 'all bachelors are not married')
The set of statements should at least in principle be falsifiable (at least one potential event exists with which the statement is consistent)
Problems
The reason why a hypothesis is falsified may be that the evidence or the preconditions are faulty, rather than the hypothesis itself. Therefore, a hypothesis is not necessarily falsified because some observation or experiment conflicts with it. The history of natural science holds significant examples of this, e.g. Newton's gravitational theory and Bohr's theory of the atom were initially falsified because of unknown factors, which were discovered later.
Popper admitted that it is often necessary to retain theories, although apparent falsifications exist. – What is left of the idea then?
Positivistic challenges Imre Lakatos (1922-1974) has further developed what he calls the 'naive critical rationalism' of Popper to a 'sophisticated critical rationalism' in an attempt to answer some of this criticism.
Positivist notions have also been challenged by Thomas Kuhn (1922-1996) in his book, "The structure of scientific revolutions" from 1962. An important point made by Kuhn is that positivist notions (including critical rationalism) do not comply with historical evidence, but that science evolves through revolutions that cause existing scientific theories to be replaced by new ones.
'Foundations of social sciences' from 1944 by Otto Neurath, points to the paradox that epistemology faces, namely the problem of circularity. Any epistemology presupposes knowledge of the conditions in which knowledge takes place, but since science cannot be used in order to ground the legitimacy of science, there are no secure foundations from which we can begin any consideration of our knowledge of knowledge. From this perspective, what we have are rather competing philosophical assumptions.
The conclusion, when looking at the results of physics and the application of scientific methods, is that most of the episodes in history that are seen as major advancements, e.g. the innovations of Galileo, Newton, Darwin and Einstein, do not correspond to the standard philosophical accounts of science.
Copy the following code to your webpage or blog to embed this presentation:
<a href="http://www.slidefinder.net/p/phd_dag_presentation_20anders_20haug/phd_dag_presentation_v1andershaug/21467044" class="slidefinder">Phd_dag_Presentation_v1 Anders Haug</a>
Comments