GIS RESEARCH UK: 6th National Conference

31st March - 2nd April 1998, University of Edinburgh, UK.

ROUND TABLE DISCUSSION CONCLUSIONS

• Historically, the performance of OODBMS (OO's) compared with RDBMS has been a problem. It remains difficult to compare the two approaches, and there are few adequate metrics or benchmarks for comparison.
• The functionality of OO [which is flexible, consistent & simple] may more than compensate for the disadvantages in performance.
• It is suggested that OO should be easier for non-specialist users to understand, although little evidence exists to support this claim.
• Re-use of software & design is a perceived advantage of OO's, but one which in practice may rarely occur.
• It would be worth comparing a range of practical applications using OO's & RDBMS's to quantify the suggested advantages of OO's in the broad arena of GIS.

• Cognitive issues : there is a need to understand the role of basic spatial data models versus photo realism. New data models are required, as well as moderation / standardisation
• Access : There is a need to address the issues surrounding freedom of information, cost and copyright. Particular application in the WWW arena.
• Public participation: There are a range of n dimensional issues - multi-scale, multi-criteria, multi-disciplinary etc, all of which are affected by ethical and power base issues of information manipulation and consumption.
• Service issues: Can GIS really act as a source of transparent management, planning and information to the Public? Control to what extent?
• Elitist technology : There is a need to eradicate the perception of GIS as a weapon and familiarise the public with its potential.

• Assuming that there is a GIS 'profession', there needs to be a more visible adherence to a code of ethics.
• Given that GIS specialists respond to the prescribed needs of others, it may be suggested that it has not, as yet, settled as a profession in its own right.
• The acronym 'GIS' may well be substituted in the future by a more wide-ranging concept GI Science of simply GI - geographic information. Interestingly - there has been little change in the core concepts of GIS curricula over the last 10 years.
• Government dictates to industry, and industry has historically driven GIS. More direct communication between government and GIS is required, which supercedes the concept of professionalism.
• It still remains a challenge to persuade industry and the public that GIS is a viable technology, and is more than just glorified IT.

• As a fundamental issue in GIS, the temporal component of spatial data remains unresolved.
• Spatial databases describe scenarios that often require different data types to be compatible within the temporal dimension. If this is not the case, analysis may be made incredibly difficult or precluded altogether.
• It is easier to extract temporal trends from certain spatial data types than others. This capability needs to be addressed from both a data storage and functionality point of view.
• OO techniques do allow the description of geographical features as objects, and thus to associate them with patterns of change or functions thereof.
• There are issues to be resolved surrounding the heterogeneity of spatial objects over time ... the problem of putting boundaries to uncertainty. An alternative to the basic temporal approach may be chaos theory which appears to be far more flexible.

• SDSS are more than a GIS or GIS as a concept. They must exhibit intrinsic GIS functionality, but in addition, demonstrate an ability to integrate geospatial data with other data formats that goes beyond current GIS application capability.
• The expert will always be somewhere. The question is where? At what point is the expert considered, consulted or ignored?
• SDSS are subject to similar constraints as GIS - eg. completeness, source, accuracy and dependability of the information. Quality remains an overall concern.
• SDSS also includes :
• risk assessment
• data quality
• confidence in output
• human/qualitative data issues

• GIS & the NHS are still relatively unacquainted.
• There are obvious and widely differing experiences between these two key players in accessing health event data.
• The issue of spatial data awareness needs to be addressed in the NHS community.
• It remains a question as to which arena it is relevant to publish material in.
• With no NHS GIS policy, it is difficult to predict or determine in what manner these two fields may co-exist in the future.

• It was agreed that it still makes sense to talk about scale. This is because:
  1. Source scale is still important
  2. Scale is still functionally relevant to specific operations.
• Questions still remain as to how the 'best' scale may be determined
• Applying the correct scale to spatial patterns was addressed particularly, as there is no standardised format for scale:pattern selection.
• Independent issues remain relevant for different scale issues.
• Global data and events have different scale dynamics to those on a more local level.
• How are scale dynamics at different levels to be integrated, and how are suitable links to be made between them?

• Selection of the technical application still depends on both functional requirement and technical expertise available.
• Distributed processing of data holds significant possibilities for advances in this field.
• Metadata has a role to play - detail and location of the spatial data needs to be considered when accessing, downloading and processing.
• If data and metadata are to be distributed, we need to be concerned about maintaining their currency and integrity as well as other DBMS issues.
• The client - server relationship needs to be enhanced beyond the distribution of data & metadata. For example, how data should be symbolised is an issue and whether a data standard needs to be adopted.

• Q. Where are we now ?
• A. Not very much further than GISRUK '97 ... or even DTM symposium '78
• GIS is still limited in incorporation of topographic features, data models, poor algorithms eg. Visibility analysis etc
• Q. Where are we going ?
• A. GISRUK '99
• Need to raise awareness of inaccuracies in data & algorithms.
• Need better terrain descriptors
• Need better data models which can work at multiple scales
• There is suggestion that 'GIS as a tool for DTM'ing should be driven by the needs of (us) ... the modellers!'