| Robert Amor's Publications in 2006 |
|---|
 |
Ma, H., Amor, R. and Tempero, E. (2006) Usage Patterns of the Java Standard API, Proceedings of APSEC06, Bangalore, India, 6-8 December, 342-349. |
Abstract: The Java Standard API has grown enormously since Java’s beginnings, now consisting of over 3,000 classes and 20,000 methods. The intent of this API is to provide high quality components that can be easily reused and so increase the Java developer’s productivity — but does it? In this paper, we present a study that begins to answer this question. Specifically we take a corpus-based approach to help determine the “typical” usage of the Standard API. We find that, in an extensive corpus of open-source software, only about 50% of the classes in the Standard API are used at all, and around 21% of the methods are used. We discuss the implications this has for future development of both the API itself, and for tools to support the API.
|
Schultz, C.P.L., Clephane, T.R., Guesgen, H.W. and Amor, R. (2006) Utilisation of Qualitative Spatial Reasoning in Geographic Information Systems, Proceedings of the 12th International Symposium on Spatial Data Handling, Vienna, Austria, 10-14 July, 27-42. |
Abstract: Spatial reasoning is a fundamental part of human cognition, playing an important role in structuring our activities and relationships with the physical world. A substantial body of spatial data is now available. In order to make effective use of this large quantity of data, the focus of GIS tools must shift towards helping a user derive relevant, high quality information from the data available. Standard GIS tools have lacked focus in this area, with querying capabilities being limited, and requiring a user to have specialised knowledge in areas such as set theory, or Structured Query Language (SQL). A fundamental issue in standard GIS is that, by relying entirely on numerical methods when working with spatial data, vagueness and imprecision can not be handled. Alternatively, qualitative methods for working with spatial data have been developed to address some key limitations in other standard numerical systems. TreeSap is a GIS application that applies qualitative reasoning, with a strong emphasis on providing a user with powerful and intuitive query support. TreeSap’s query interface is presented, along with visualisation strategies that address the issue of conveying complex qualitative information to a user. The notion of a relative feature is introduced as an alternative approach to representing spatial information.
|
Schultz, C.P.L., Guesgen, H.W. and Amor, R. (2006) Computer-Human Interaction Issues when Integrating Qualitative Spatial Reasoning into Geographic Information Systems, Proceedings of CHINZ’06, Christchurch, New Zealand, 6-7 July, 43-51. |
Abstract: To allow the immense volume of spatial data currently available to be used effectively, people need intelligent query tools that are simple and intuitive. Standard query tools have a number of serious usability limitations, as they often rely solely on numerical approaches when dealing with spatial information. The qualitative reasoning community has addressed this issue, by providing powerful formalisms based on the way that humans deal with spatial information, however, integrating these methods into numerical systems raises a number of new CHI problems. This paper addresses three key CHI challenges when combining qualitative and numerical methods: (1) managing the subjective, ambiguous nature of qualitative terms, (2) providing a powerful, yet simple query system, and (3) effectively visualising a complex, fuzzy qualitative query solution. A qualitative GIS called TreeSap is presented, which demonstrates that, with the use of CHI principles, query tools can be both powerful and accessible to non-expert users.
|
Ma, H., Ha, K.M.E., Chung, C.K.J. and Amor, R. (2006) Testing Semantic Interoperability, Proceedings of the JICCDMCBE, Montreal, Canada, 14-16 June. |
Abstract: With standardised semantic representations of construction objects able to be transferred between major CAD systems, and other design tools, there is an expectation, supported by compliance testing, that semantically consistent data will flow across the project team. This assumption is questioned due to the known difficulties in mapping consistently and completely between two distinct representations of an artifact. To test the ability of CAD, and design tools, a number of buildings, described in a standard format, are loaded and then saved directly back out of these tools and then checked for differences. A range of potential differences has been postulated, and experiments show the existence of most categories of differences when data files are examined.
|
Plimmer, B. and Amor, R. (2006) Peer Teaching Extends HCI Learning, Proceedings of ITiCSE 2006, Bologna, Italy, 26-28 June, 53-57. |
Abstract: Crafting a good user experience requires skills in several disciplines. Few people have this breadth of knowledge, and undergraduate computer science students are no exception. Encouraging computer science students to appreciate the ways that other disciplines contribute to Human Computer Interaction is important, yet difficult. Our students learn about this disciplinary interdependence through peer teaching as part of a group project. Each group contains students with complementary skills and we expect a transfer of knowledge. Here we discuss the educational theory behind the project, the project’s essential elements and an evaluation of how it aids learning. The model we have developed could be easily adapted for other courses which draw on diverse skills.
|
Lin, M-Y., Amor, R. and Tempero, E. (2006) A Java Reuse Repository for Eclipse using LSI, Proceedings of ASWEC 2006, IEEE Computer Society, Sydney, Australia, 18-21 April, 351-360. |
Abstract: Software Reuse is a concept that is frequently mentioned as a way to improve software developers’ productivity. However, there are a number of issues that need to be addressed in order for software reuse to be adopted by developers. One of those issues is providing enough reusable artifacts. The Java Standard API has been quite successful in this, with the latest version having over 3000 classes available. However this raises the issue of finding the right artifact to reuse. With the Java API, this means trawling through the JavaDoc webpages, which has the risk of not being able to find the right artifact, even though it is in the API. In this paper, we explore the use of latent semantic indexing as a means to index the Java API JavaDoc pages. Specifically, we describe Prophecy, an Eclipse plug-in that presents the Java API as a software repository.
|
Amor, R. and Ma, H. (2006) Preservation of Meaning in Mapped IFCs, Proceedings of EC-PPM 2006, Valencia, Spain, 13-15 September, 233-236. |
Abstract: The development of high-level standard representations of buildings has been welcomed by the majority within the industry. Such standards allow for semantic interoperability between the large number of design tools which are available to practitioners in the A/E/C and FM industries. However, with such standards comes greater reliance on the information contained within the models which are transmitted, and interpreted, automatically between design tools. Unlike geometry-based standards (e.g., DXF, IGES, DWG, etc) where there was always an expectation of human interpretation, this semantic data must be correct to fit within the interoperable world we have developed. Testing of the semantic interoperability of a small number of commercial design tools has shown that this level of trust is not yet able to be assumed and that further work needs to be done to ensure that we can preserve meaning when moving semantic information between design tools.
| O’Sullivan, D., Clephane, T. and Amor, R. (2006) Integrating agent models and GIS: towards an extensible open source solution, Presented at International Geographical Union 2006: Regional Responses to Global Changes - A View from the Antipodes, Queensland University of Technology, Brisbane, Australia, 3-7 July. |
Abstract: Agent-based models are widely used to study geographical phenomena as diverse as the diffusion of disease, or the impact of sprawl on emergency management. From a geographical perspective, a key failing of agent-based models is that most run in grid-based abstract spaces, and such simplified topologies only approximate the complex spatial interaction structures of real geography. Richer geographical representations are required if realistic models are to be developed. As Gilbert and Bankes (2002, 7197) point out, existing agent-based modeling toolkits, while “[…] excellent for simulations involving agents located on a rectilinear grid [...] are less useful for simulations [...of...] actual terrain.” (Gilbert and Bankes 2002, 7197). We describe initial progress integrating an open source agent-based modelling toolkit (Repast) with an open source GIS API (GeoTools), using the adapter and command programming patterns. Each toolkit has an adapter that allows it to react to commands from a controller, routed via a registry mechanism. Building, initializing and running a model involves issuing commands to the two toolkits. This architecture offers three advantages: (i) it removes direct dependencies between toolkits, allowing their future modification or replacement by alternative toolkits, when only different adapters would be required; (ii) the command pattern enables future development of a declarative modelling language, since any model based on the toolkits can be built by a series of commands; (iii) the potential is provided to link a model via new adapters to further toolkits (for example, for geovisualization). A significant remaining challenge is the development of a standard 'spatial agent' which would be required for a fully extensible system. At present users must define agents for models by programming them in Java.