Saturday, May 30, 2009

Collaboration in Engineering using Semantic Web techniques

Abstract
This article describes how software developers could enable domain experts such as engineers to collaborate on problem solving. The means for this is development using Semantic Web and Web 2.0 techniques to enable end-user programming. This could enable domain experts such as engineers to be involved in the modelling of a problem such as product design, and so understand, assess, and develop possible solutions.

Introduction
Visualisation is necessary to provide traceability for decision making, as an audit trail for information, to justify a decision and/or cost, or to understand a process, or a product data structure. It is necessary to navigate a database in order to validate it, the structure of the visual interface needs to show the structure of the data, and the model. The best way to do this is to relate them all so that any change to the database, changes the structure of the model and this changes the structure of the visualisation, or vice versa . Semantic search allows someone to see results or individual items but enables going straight to this, that doesn’t show the context and the model is still a ‘black box’, unless the Semantic search actually returns the relevant portion of the structure so that the context of the returned information is displayed.

Methodology
The example below illustrates how it is possible to enable refining a search by visualising all the items present in sub-categories of the main category found in the search. McGuinness (2003) explains how ontologies support this functionality and calls this ‘generalization/specialization’ of information. Uschold and Gruninger (2004) describe ‘ontology-based search’ as “Ontology used for concept-based structuring of information in a repository”; and describe the benefit of this as “better information access” This aids the research objectives of ease of use and sharing of information. The use of open standards for representing information makes it possible to enable searches that understand the semantics of the information and so can track all of the relationships between items. This example illustrates the interface for making a search. In this example the user wants to retrieve all the information related to a spar (aircraft wing component).





Semantic Search interface


The result is shown as a series of trees for each item that contains the word spar. Each keyword match is the root of a tree. Each tree shows the item found and all its children and attributes. The example below shows an image of the top part of the results, this is part of the branch for the first item returned.



Results from semantic search

The information is held in linked and related taxonomies/sub ontologies so it is not HTML that is being searched but the taxonomy itself. Because the information is held in a structured way, it is much more likely that searchers will find what they are looking for, because the search can follow the relationships represented in the taxonomy. One of the key objectives of Semantic Web research and Web 2.0 is to make this kind of search possible over the web as a whole. The Semantic Web is a longer-term vision for managing information over the web and Web 2.0 is the shorter-term practical implementation of techniques, which can ease current information search and management problems. A web interface has been developed for Protégé An example of the use of this is the illustration below where a search is made for information on the cure cycle for composites manufacturing. This search is possible as WebProtege[1] has succeeded in providing a web based interface for displaying and searching ontologies, so providing an additional way to enable web access to the test ontologies created for this research.



Web Protégé Interface


The next stage of this problem is to enable interaction and modelling with the returned information. A project was created called Bitriple by Leaver (2008), to enable end-user functionality for this kind of web-based ontology construction and search. The application provides a facility to edit an ontology/ies and instances, and provides tree-based visualisation of the ontology (as shown in the example below). This example illustrates creation of an online aircraft wing ontology. Wing component sub/ontologies created using Protégé can be translated for the Bitriple application to be represented as RDF/XML. An application could be built as an extension to Bitriple to perform calculations and modelling using the information stored. This could assist in allowing domain expert end-user programmers/modellers to create models. Such web applications provide an alternative to spreadsheets, and to single computer based programs; and if installed on a network server, such applications can provide a collaborative model development environment. This would fit in well with both Semantic Web, and Web 2.0 approaches to knowledge creation, allowing structured representation and navigation of information, and end-user interaction, collaboration, customisation and programming via the web. Collaboration can aid people to agree on terminology, and standardisation of calculations used such as for cost rates and currencies. RDF information can be searched with SPARQL[2] (SPARQL Protocol And RDF Query Language), which is used to search the Bitriple application.

A screenshot from the Bitriple application, of ontology creation for an aircraft wing, is shown below :-



Bitriple Ontology Creation Screenshot


Conclusion

If domain experts are not involved in creation of the model, there is no proper way for the knowledge they are encoding to emerge, or for proper collaboration so each expert is best encapsulating the part of the model that they are most expert in. People need to be involved in model creation if they are to understand the model, decision, and how the decision was made. So there is a need for end-user modelling/programming to enable this. Full automation hinders user involvement and traceability, so semi-automated systems that interact with end-users and assist with all stages of the model decision are better. If a person goes straight to the answer how can it be expected that they fully understood the question?

Breaking complex mathematics into modularised traceable steps eases management of it, and visualisation, and allows modelling of different scenarios, and these scenarios demonstrate the emergent properties of the model, enabling decision support. BiDirectional Traceability is needed, traceability between nodes/submodels, and between models, and between suppliers and buyers.

Existing ontologies can be extended for modelling of software systems and engineering systems e.g. PSL, STEPML, UML, SysML can be extended.This research can enable collaborative modelling and interaction, via applying end-user programming techniques to enable domain experts such as engineers and business people to create and interact with the knowledge representation themselves, and co-operate to ensure the representations are useful for addressing their problems, with less software creation barriers. Software developers need to enable such systems to make this all possible.

[1] SPARQL is the query language and protocol for RDF being recommended to the World Wide Web Consortium - (2008) [online]. Available from: http://www.w3.org/TR/rdf-schema/ [Accessed 27 June 2008]. A tutorial has been developed by Dodds - [online]. Available from: http://www.xml.com/lpt/a/2005/11/16/introducing-sparql-querying-semantic-web-tutorial.html
[Accessed 27 June 2008].

[2] WebProtege (2008) Wiki Page [online]. Available from: http://protege.cim3.net/cgi-bin/wiki.pl?WebProtege [Accessed 27 June 2008].
References
Leaver, N. (2008) Using RDF as an Enabling Technology. MSc. Dissertation, University of the West of England, Bristol.

McGuinness, D. L., 2003. Ontologies Come of Age. In: Dieter Fensel, Jim Hendler, Henry Lieberman, and Wolfgang Wahlster, ed. Spinning the Semantic Web: Bringing the World Wide Web to Its Full Potential. MIT Press, 2003.

Uschold, M., Gruninger, M., 2004. Ontologies and Semantics for Seamless Connectivity. In: Association for Computer Machinery - Special Interest Group on Management of Data - SIGMOD Record December, 33(4).

Monday, May 25, 2009

Enabling diagrammatic de-abstraction and modelling of engineering problems - Summary

This post discusses efforts to enable de-abstraction of engineering problems from a representation suitable for engineers to that suitable for computer models and code. The key question is to what extent diagrammatic representations of problems can be used in order to provide modelling solutions. To achieve this, a source tree is created, this is translated to computer code, then represented as a result tree. This enables engineers to visualise problems such as representation of a product data structure in a way familiar to them, and this also gives a visual and colour coded representation of equations. This visualisation is easier to navigate and understand than that which can be provided by spreadsheets, and more maintainable. This research could also be used for business modelling, process modelling, and workflow.

The essence of the problem is that organisations and individuals have many limitations resulting from the lack of facilities to allow users to program. Organisations have many limitations resulting from the lack of facilities to allow users to program. For example the use of 'out of the box' modelling tools, which are hard to customise or introduce collaborative capabilities, because a project deadline is so urgent that nothing else is practical. The problem with this is that when organisations have urgent needs to produce models for a particular deadline, all their effort goes into this. Once a deadline is met or missed, there will be a new deadline for something else. So all the effort goes into meeting these deadlines, and there are little or no resources for providing a better solution that makes it easier in the long run to meet the deadlines. This causes many issues, which need to be urgently tackled by organisations in the short term. This creates an ironic dilemma, that the problems caused by lack of end user programming facilities cause there to be a need to tackle these problems via quick fixes because they are so urgent. This makes it difficult to secure funding and time for longer term end user programming research.

As an example of the problems that occur; many people use Excel for financial modelling, and engineering, business, and science models. As these models become large, the short term solution is to use Excel Visual Basic to aid navigation through the model. However, though this helps with the immediate problem, in the longer term it can be difficult to reuse the model, as the Excel spreadsheet creator has to go through his or her code and remember what they have done, this problem is worse if a different user is trying to reuse the spreadsheet. Experience of dealing with projects that create new software systems is that they are very ambitious, but not very modular or customisable. Under management pressure for meeting of timescales, they are often released too early with insufficient consultation with users. Then software developer's time is reallocated to a help desk, in order to deal with the problems caused by the release of software too early, too unchangeable and with too little user involvement.

A longer term solution would be to provide a visual programming environment where the structure of the code is visualised, to make it easier to track the formulae, and so navigation is a central part of the program design, this also minimises the need for code. A solution is to produce highly customisable software, so that the software team do not have to anticipate every problem the users will want to solve. Then many users with more advanced needs and computing abilities could customise software for their needs.

Engineering modelling can be performed using a high level diagrammatic view of the problem and conveyed to the computer via transformation. Program/model transformation allows for writing in one representation or language, and translating to another. This is particularly useful for language independent programming, or for high level and end user programming that can then be translated to a language more easily interpreted by computer systems.

Even if programming is made easier, only a proportion of people would actually be interested or capable of doing this. But there is still an advantage to colleagues such as people in the same team or department as an end user programmer, even if they are not undertaking programming themselves. Then all in the team have much closer access to someone, the end user programmer, who understands their, and the team's tasks, requirements, and projects. This closes the gap between those producing software systems, and those who require the software. This also makes it easier to iterate through solutions and solve problems more quickly and collaboratively.

Application of this kind of research at Rolls-Royce are described by Scanlan et al. (2006). Collaboration, simulation and modelling have been investigated to determine the requirements for future research in modelling of problems. This should allow translation from a model based representation of software to the actual software. This can involve semi automatically producing software for a Semantic website from visual representations of the problem. The core of this modelling infrastructure is automated generation of models created with World Wide Web Consortium (W3C) standards based languages, and the visualisation of information represented in such W3C standard ways. This research investigated alternative approaches to software development, which give users greater involvement, and can actually be used in combination. This partially automates the process of software creation via a collaborative process and equation tree that maps the problem structure, and user interface creation by providing a means to manage a diagrammatic and/or tree based representation.

Reference

Scanlan, J., Rao, A., Bru, C., Hale, P., Marsh, R., 2006. DATUM Project: Cost Estimating Environment for Support of Aerospace Design Decision Making. Journal of Aircraft, 43(4).

Wednesday, May 20, 2009

Enabling diagrammatic de-abstraction and modelling of engineering problems

Abstract
This paper discusses efforts to enable de-abstraction of engineering problems from a representation suitable for engineers to that suitable for computer models and code. The key question is to what extent diagrammatic representations of problems can be used in order to provide modelling solutions. To achieve this, a source tree is created, this is translated to computer code, then represented as a result tree. This enables engineers to visualise problems such as representation of a product data structure in a way familiar to them, and this also gives a visual and colour coded representation of equations. This visualisation is easier to navigate and understand than that which can be provided by spreadsheets, and more maintainable. This research could also be used for business modelling, process modelling, and workflow.

Introduction
C.S. Peirce (1906) stated in 'Prolegomena to an Apology for Pragmaticism' "Come on, my Reader, and let us construct a diagram to illustrate the general course of thought; I mean a system of diagrammatization by means of which any course of thought can be represented with exactitude". That is the purpose of this research, but to limit the scope and so make application of this theory testable the research is restricted mainly to engineers (because they often think in terms of diagrams) and to the domain of modelling (which often requires diagrams). So the aim is to apply the research first where it can have the most use and encourage others to expand it for other domains and other users. This research is intended to simplify computing for computer literate non-programmers, this includes many engineers. The main research area is enabling users such as engineers to model the problems they encounter in manufacturing and design. However, the wider aim is to prototype research for enabling a larger range of software users to model their problems. The intention is to create collaborative tools that allow users to develop software in a way they will be familiar with from their use of spreadsheets. This research brings together approaches of object orientation, the Semantic Web, relational databases, and Model-Driven and Event-Driven programming. Frankel et al. (2004) explain the opportunities for, and importance of this kind of research.

Figure 1 shows the way iterative development is used both in this research and in the implementation to ensure that changes can be made systematically as necessary and without disrupting the project.


Figure 1. Research Development Iterations

Software engineering and modelling has much in common with engineering modelling, also the tools used for both have much in common. Software process modelling, engineering process modelling, and business/workflow modelling share a common approach, and similar tools. Much of this commonality is in the need to transform requirements into design into code semi-automatically. To achieve this, continuous consultation between potential users e.g. engineers for engineering modelling problems and developers for software problems is required.

Methodology
A common factor in these various types of modelling is the need to transform between a high level abstraction, to a lower level such as a computer model and then code. This is illustrated by examples of semi-automatically produced programs/models (Hale, 2008). The translation process involves translating from a tree/graph representation and for each node this is translated into a code representation of the equation that relates this node to any others, and this code is then presented in the interface as a result tree/graph. This can be achieved for programs and/or web pages. Kraus et al. (2007) examine and implement this transformation problem and also produce code and/or web pages. Uschold (2003) defines the Semantic Web as being machine usable and associated with more meaning. So this is a good way to convey the abstractions represented in a source and result tree to the end user.

The intention is to demonstrate a way to construct diagrammatic representations of cost using the example of an aircraft wingbox. The wingbox is the structure or skeleton of the wing. These diagrammatic representations are achieved by visual representation of items and equations that make up wingbox cost. These items and equations can be represented in standardised categories used in engineering - ‘materials’, ‘processes’, ‘cost rates’ etc. These categories are standard for engineering and the methods for representing items and equations that relate the items can be expressed in standard mathematical form. Therefore using the same methodology and same categories it would be possible to represent other items and equations in the same way. So this methodology is reusable for costing other engineering components including those outside aerospace. The costing method is also recursive because components and sub components can be costed separately or together and top down or from bottom up. This methodology has the potential to be applied to any calculation based modelling problem.

Solutions to this transformation problem can be found by adapting current tools and techniques using a systematic approach. Such tools and techniques involve use of modelling tools, spreadsheets, ontology management tools, and Semantic Web Web 2.0 tools. These possible solutions are not mutually exclusive and their combination could be the best way of providing usable collaborative modelling tools for computer literate end users and domain experts. The link between these alternative ways of advancing current research is translation and User Driven Modelling/Programming.

The User Driven Modelling/Programming approach advocated in this thesis has the advantage that it is using a modelling approach for creating modelling solutions and involves creating systems to create systems. This makes it possible to solve the problem by breaking it down into stages and allowing software developers to concentrate on the most complex software problems and domain experts to be able to concentrate on their domain problem. The standardisation possible in this approach can allow software developers to create modelling systems for generic purposes that can be customised and developed by domain experts to model their domain. This methodology can be facilitated by :-

· Modelling Tools - Building an end-user interface and extending the translation capabilities of UML (Unified Modelling Language) and/or other modelling tools (Johnson, 2004).
· Spreadsheets - Improving the structuring and collaboration capabilities of spreadsheets, and enabling customisation of spreadsheet templates for particular domains and users.
· Ontology Tools - Extending the modelling capabilities and equation calculations in ontology tools and providing an end-user interface.
· Semantic Web/Web 2.0 - Extending the capabilities of Semantic Web and Web 2.0 style web based development tools to allow collaborative modelling.

Figure 2 shows the solutions, and how these make User Driven Modelling/Programming possible :-


Figure 2. Methodology Diagram - Enabling User Driven Modelling/Programming

It is possible to create an extra layer of visualised semantics to enable users to specify commands in structured language. This approach of adding extra layers is the way this visual programming works. Users provide the information the program needs at the visual interface layer, and program code is created automatically. The layers provide the bridge between abstract ideas and computer code. If this approach is taken to its logical conclusion, it would be possible to allow the user to specify what the computer should do. Then each layer would communicate this to the layer below until the computer performs the action required. A simple example of this approach is the use of spreadsheets. Users can specify a calculation in mathematical terms using a formula. The spreadsheet then calculates the result of the formula. Users can change the formula if it is incorrect without any need to write code or re-compile. This accounts for the popularity of spreadsheets. However, spreadsheets do not provide the centralised and structured data-store required for a distributed collaborative system. Therefore, the research concentrates on combining the wide applicability of generic spreadsheet modelling with structured and adaptable modelling and visualisation.

It is important to enable changes to the design of the information source and its structure as necessary, even when it contains information. This makes possible continuous improvement of the information and its representation together. Clear visualisation of the structure makes out of date and duplicate information obvious, so it can be changed by the end-users of the information. This provides for maintenance of information quality without necessitating end-users to understand relational database design; though relational databases can still be accessed by software specialists for more in depth and less frequent structural changes.

Program transformation allows for writing in one representation or language, and translating to another. This is particularly useful for language independent programming, or for high level and end user programming that can then be translated to a language more easily interpreted by computer systems.

A taxonomy representation is translated into a computer model. Relationships can be conveyed to a software model that evaluates them. Information is translated from the taxonomy and is visualised in tree form in a decision support tool with the example of spar manufacture information. The visualisation of the information in a tree can be further translated into visualisation as an interactive diagram. The representation can be translated into different languages, to allow for language independence.

Figure 3 explains the transformation process.


Figure 3. Translation Process

Related Research
Crapo et al. (2002) assert the need for a methodology for creation of systems to enable more collaborative approaches to modelling by domain expert end-users, and that this combined with visualisation would allow engineers to model problems accurately. Huhns (2001) and Paternò, (2005) both explain that alternatives to the current approach to software development are required. Modelling languages such as Alloy explained by Wallace (2003) can be used as an interface to an End-User Programming environment. Transformation from a model building environment to program code has been investigated by Gray et al. (2004).

Conclusion
Experienced programmers can build a modelling environment that can then be used by non programmers to create models or solve other software problems. This was achieved for the DATUM (Design Analysis Tool for Unit-cost Modelling) project with Rolls-Royce, and the modelling environment created was used by their engineers. This is described by Scanlan et al. (2006). Collaboration, simulation and modelling have been investigated to determine the requirements for future research in modelling of problems. This should allow translation from a model-based representation of software to the actual software. This can involve automatically producing software for a Semantic website from visual representations of the problem. The core of this modelling infrastructure is automated generation of models created with World Wide Web Consortium (W3C) standards based languages, and the visualisation of information represented in such W3C standard ways. This research investigated alternative approaches to software development, which give users greater involvement. This partially automates the process of software creation via a collaborative process and equation tree that maps the problem structure, and user interface creation by providing a means to manage a hypermedia concept map.

References
Crapo, A. W., Waisel, L. B., Wallace, W. A., Willemain, T. R., 2002. Visualization and Modelling for Intelligent Systems. In: C. T. Leondes, ed. Intelligent Systems: Technology and Applications, Volume I Implementation Techniques, 2002 pp 53-85.

Frankel, D., Hayes, P., Kendall, E., McGuinness, D., 2004. The Model Driven Semantic Web. In: 1st International Workshop on the Model-Driven Semantic Web (MDSW2004) Enabling Knowledge Representation and MDA® Technologies to Work Together.

Gray, J., Zhang, J., Lin, Y., Roychoudhury, S., Wu, H., Sudarsan, R., Gokhale, A., Neema, S., Shi, F., Bapty, T., 2004. Model-Driven Program Transformation of a Large Avionics Framework. In: Third International Conference on Generative Programming and Component Engineering GPCE, pp 361-378.

Hale, P. 2008. http://www.cems.uwe.ac.uk/~phale/EconomicModels/ModelsVisualised.htm.

Huhns, M., 2001. Interaction-Oriented Software Development. International Journal of Software Engineering and Knowledge Engineering, 11, pp 259-279.

Johnson, P., 2004. Interactions, collaborations and breakdowns. In: ACM International Conference Proceeding Series; Proceedings of the 3rd annual conference on Task models and diagrams Vol 86 Prague, Czech Republic.

Kraus, A., Knapp A., Koch, N., 2007. Model-Driven Generation of Web Applications in UWE. http://ftp.informatik.rwth-aachen.de/Publications/CEUR-WS/Vol-261/paper03.pdf In Proc. MDWE 2007 - 3rd International Workshop on Model-Driven Web Engineering, CEUR-WS/, Vol 261, July 2007.

Paternò, F., 2005. Model-based tools for pervasive usability. Interacting with Computers, 17(3), pp 291-315.

Peirce, C.S. - 1906. Prolegomena to an Apology for Pragmaticism - http://www.existentialgraphs.com/peirceoneg/prolegomena.htm.

Scanlan, J., Rao, A., Bru, C., Hale, P., Marsh, R., 2006. DATUM Project: Cost Estimating Environment for Support of Aerospace Design Decision Making. Journal of Aircraft, 43(4).

Uschold, M., 2003. Where are the semantics in the semantic web? AI Magazine Vol 24 (3) pp 25-36.

Wallace, C., 2003. Using Alloy in process modelling. Information and Software Technology, 45(15), pp 1031-1043.


Wednesday, May 13, 2009

Google unveils 'smarter search' - BBC News

This is an interesting article about Google's plans, and explains that they are keen to develop search tools that make use of the Semantic Web. So I refernce and link to this article below

By Maggie Shiels Technology Reporter, BBC News, Silicon Valley

"Web giant Google has unveiled new products that it says will push search in a new direction.

Google is using so-called semantic web technology to leverage the underlying data on websites to enhance results."

http://news.bbc.co.uk/1/hi/technology/8047076.stm.

This article is also relevant to smarter search -

Wolfram 'search engine' goes live

"A web tool hailed as a significant rival to search giant Google has gone live to the public.

Wolfram Alpha is called a computation knowledge engine rather than a search engine and wants to change the way people use online data.

It aims to give people direct answers to queries rather than send them to other sites where they may find what they are seeking.

The system is the brainchild of British-born physicist Stephen Wolfram."

http://news.bbc.co.uk/1/hi/technology/8052798.stm.

Monday, May 04, 2009

End-User Programming using Modelling and Semantic Web

Even if programming is made easier, only a proportion of people would actually be interested or capable of doing this. But, there's still an advantage to colleagues, people in the same team or department as an end user programmer, then all in the team have much closer access to someone who understands their, and the team's tasks, requirements, and projects. This closes the gap between those producing software systems, and those who require the software. This also makes it easier to iterate through solutions and solve problems more quickly and collaboratively.

Software tools to improve and combine UML and other modelling, spreadsheet, Web 2/Semantic Web, and software solutions are required to allow this kind of end-user participatory development and a sustained collaborative organised push towards achieving this. To achieve this, end users and teams need to be able to visualise their solutions both literally as diagrams, and metaphorically via problem solving collaboration. This would realise the intentions that were expressed decades ago in the form of PC based computing and spreadsheets, but add better information and modelling organisation, and collaboration capabilities.

To make the above practical sustained, research is needed in these areas. Such research would have many practical spin-offs.

This page of Professor Richard Vidgen - http://www.bath.ac.uk/management/faculty/richard_vidgen.html - describes research into "how technology is developed and used within organizations. I am interested in how software systems are constructed, e.g., development methods and agile approaches; I also study the interplay between people and technology using complex systems theory, particularly coevolutionary theory, and social network analysis. I have further interest in evaluating the quality of e-commerce offerings and continue to work in the area of web quality."

So this is relevant to my subject area.

My home page is - http://sites.google.com/site/userdrivenmodellingprogramming/.