Your report should be limited to approx. 1500 words (not including references). Use 1.5 spacing with a 12-point Times New Roman font. Though your paper will largely be based on the chosen article, y

Fusing Scientific Modelling with Knowledge Management Systems S. Kininmonth a,c, T. Donovan b and S. Edgar a a. Australian Institute of Marine Science, Townsville,, Queensland Australia b. CRC Reef, Townsville, Queensland, Australia c. Corresponding author Ph. +61-7-4753443 [email protected] Abstract: The influence of scientific problem solving on policy makers and the general public is a direct function of the perceived relevance and clarity of communication output. However environmental modelling is complex and this diminishes the ability of traditiona l media to clearly represent model output. Simplified brochures and news articles struggle to convey the complexity of the network of abiotic and biotic interdependent systems. Conversely detailed reports and scientific papers are rarely read. The optimum delivery of science involves the view er dynamically determining the level of detail for a topic of interest.

Static media such as newspapers and non-digital television have no provision for information expansion. This paper presents a knowledge management model focusing on integrating GIS into dynamic web pages to deliver a scalable information product. The web site called www.ReefFutures.org uses dynamically generated web pages (Coldfusion) to incorporate GIS (ArcIMS). Using textual search engines (Verity) within a highly graphical and dynamic website permits the viewer to fully explore a topic of interest. The interactive mapping technologies permit the viewer to zoom in and fully investigate the scientific complexities surrounding an environmental issue. For many viewers this will be the first time to examine the raw data used by scientists. The first issue be ing addressed is coral bleaching on the Great Barrier Reef. Spatial data depicting recent coral surveys can be viewed in the full context of satellite derived sea surface temperatures.

Hyperlinking offers a mechanism of connecting images of the corals from surveys to the map elements with metadata describing various modelling techniques used. The ability to deliver complex modelling results through a sophisticat ed communication delivery mechanism will significantly enhance the influence of science. Keywords: science communication, knowledge management; dynamic web page, webmap, coral bleaching 1. INTRODUCTION The transition of knowledge generated by scientific modelling to knowledge consumed by society is often ineffective and cumbersome. Yet most scientists are driven by a passion to influence the society in which they live. This passion is particularly prominent for natural resource scientists who study threaten environments. Ironica lly the field of environmental modelling is often hidden from environment-dependant society due to research complexity and ambiguity. To influence the policy makers and the general public requires a high degree of perceived relevance and clarity of communication output (Freyfogle and Newton 2002). Traditional media, with concise and targeted articles, struggle to convey the complexity of the intertwined network of abiotic and biotic interdependent variables. Formal scientific publications such as papers and reports serve to provide credence and relevance to the scientific community but are rarely read by the wider society. Th e optimum delivery of science involves the viewer dynamically determining the level of detail for a topic of interest. Static media such as newspapers and non-digital television have minimal provision for information expansion. Complex issues, such as coral bleaching, are represented as simplistic articles like “Bleaching ha s reef in hot water” (The Australian, 20 th August 2001, page 7). Knowledge Management Systems (KMS) offer the potential to convey complex research results to a wider audience. The definition of KMS is broad and confused (Hlupic et al. 2002) but can be defined as “technologies which enhance and enable knowledge generation, codification and transfer”(Hlupic et al. 2002) p.91). Knowledge and information are often confused and this directly relates to the complicated and non-linear creation of knowledge from information (Styhre 2002). Information is data that creates a change in the receiver’s knowledge and then becomes obsolete or non-information (Styhre 2002). As Styhre (2002 p.230) states the repetition of non- informative data, such as advertising, that makes no difference “may even be frustrating or somewhat annoying”. Knowledge, however, is not the accumulation of information but rather the intelligent use of information (Hlupic et al. 2002, Styhre 2002). To complicate matters further, only knowledge relevant to a specific decision making process is desirable, and designing filters is a fundamental problem for KMS (Fink 2002). To be effective, KMS requires the strategic implementation of knowledge management tools.

Hlupic et al (2002, p.95) outline three categories of tools that can be utilised within a KMS: • “Knowledge Generation – the creation of new ideas, recognition of new patterns, the synthesis of separate disciplines, and the development of new processes.

• Knowledge Codification – the auditing and categorisation of knowledge.

• Knowledge Transfer – the forwarding of knowledge between individuals, departments and organisations.” For many scientists the tools of knowledge generation (ie statistical modelling, data mining) have been the focus at the expense of knowledge transfer tools. Work programs at research institutes often restrict human resource capacity to scientists with knowledge generation skills with scientific papers being the preferred and limited knowledge transfer tool. To address a need for wider communication, libraries now focus beyond codification to metadata management with search engines and online retrieval systems that facilitate knowledge asset management (Williamson and Liopoulos 2001). Digesting formalised knowledge is often tedious and science communication staff are used to assist re- engineering of scientific papers into popular media. Using journalistic techniques these staff are able to extract clarity and relevance from research modelling. The World Wide Web provides a rapid and convenient means of knowledge dissemination (Boechler 2001). To display and arrange condensed articles on the web requires web site managers. In many cases this system of scientists, librarians, communicators and web site managers is fully functional and satisfies the clients needs. For example the Long Term Monitoring Program at Australian Institute of Marine Science has developed a web site ( http://www.aims.gov.au/pages/research/reef- monitoring/reef-monitoring-index.html) that facilitates rapid transfer of monitoring information with derived knowledge concerning the ecological status for the Great Barrier Reef (Sweatman et al. 2001). New technologies in web site design and function have created a new array of transfer tools (Fischer 2001). In particular the use of interactive web-based mapping ( webmaps) has added a spatial dimension to the textual and static picture displays of the past. The spatial dimension is the prime focus of Geographic Information Systems (GIS) and has been the domain of speci alists since inception (Talen 2000). With the design of internet mapping servers the web viewer has the ability to directly request map images compos ed in real time from stored data. This data can be the actual data used by the scientists in their modelling systems.

Although presently limited in analytical functionality the spatial servers do offer query and display functionality. The simple ability to view collateral datasets for an area of interest at a suitable scale is particularly powerful. Interactive maps by themselves do not have sufficient context to be valuable in knowledge transfer (Boechler 2001). Dynamic web pages with search capability interlinked to the webmaps provide knowledge transfer capacity.

The principal technological development has been the design of web application servers. Contrary to web servers, the application servers are able to interact with database s, deliver customised information on user preferences and validate user actions (Macromedia 2002). This paper describes the KMS at the Australian Institute of Marine Science with particular emphasis on the knowledge transfer tools. The first section will outline the knowledge generation and codification tools across the institute while the second section will focus on delivering a knowledge transfer tool th at is targeted towards disseminating coral bleaching research.

2. SYSTEM ARCHITECTURE The Australian Institute of Marine Science (AIMS) “was established by the Australian Commonwealth government in 1972 to generate the knowledge needed for the sustainable use and protection of the marine environment through innovative, world-class scientific and technological research” (www.aims.gov.au/pages/about.html). Thirty years of accumulated knowledge reside within the institute and this requires management to optimise future knowledge creation. Issues of integration across research groups combined with data quality and accessibility have highlighted the need for a more sophisticated system (Kininmonth 2002). A KMS is under construction and consists of three integrated elements (figure 1); a n Ente rprise Ge ograp hic In form ation Sy ste m (EGI S), a data cen tre (ADC) an d kno wled ge trans fer servers (W eb serv ers) . 2.1 Im plementation Th e EGIS is a clien t fo cused GIS th at utilises th e fun ction ality d eliv ered by ESRI’s su ite o f software utilities. Fig ure 2 pro vides a diagrammatic overview of th e system with in AIMS. Pi votal to th e su ccess of EGIS is th e cen tralisatio n o f qu ality assu red sp atial d ata combined wi th stan dardised m etad ata an d multiple acces s opportunities (Kini nmonth 2002). Sp atial d ata with in AIM S is sto red in a multi-u ser geo databa se u sing E SRI’s Spat ial Dat a En gine (SD E). St rict dat a-nam ing co nve ntions are em ploy ed t o keep t he dat a layers o rganised, whi le a sy stem of u ser pri vileges allows fine-grained cont rol of clients accessi ng the data . Se curity con siderat ions asi de, the geodat abase m odel ha s furthe r adva ntages in spee d of data access (Zeiler 1999). La rge -scale im ages are store d as sea mless raster d atasets, an d data retriev als are automatical ly co nstrain ed to th e clien t's map extent. Im age pyram ids - a pre-calc ulated resam pling of the im ages - are used to furthe r im pro ve dr awi ng per formance. Vect or dat a is also extrem ely fast, as the geom etry is store d intern ally, and an in tern al syste m of grid-b ased inde xing tables are use d in spatial queri es. Figu re 1. T he Kn owledge M anagem ent Sy stem at AIM S wi th the ge nerat ion an d c odi ficat ion syste ms e mbedded wit hin the scien tific m odellin g envi ronm ent. The tra nsfe r syste ms create a link bet ween t his envi ronm ent an d t he wi der world. Access is through a num ber of interfaces includi ng t he ArcI nfo s oftware (t hick c lient), webm aps and cust om ised Java application programming interfaces (thin clients ). The webm aps are service d by the Arc Internet Mappi ng Se rvice (A rcIM S) w hich p rovi des access through a common we b browse r. Im portan tly the m etad ata th at describ es the spatial data is also served via a web interface powe red by Arc IMS. R esearche rs can blend a nd m anipul ate data to su it t heir m odelling requ irem ents. Th e EGIS is esp ecially criti cal at A IMS wh ere d ata stora ge excee ds several terabytes and dat a mining techniques a re require d (Koperski et al. in press ). Th is system will g reatly aid th e know ledg e generation too ls cu rren tly b eing utilised (see Wo oldridge a nd Do ne, this edition). Figu re 2. The struct ure of EG IS The AIMS dat a centre (ADC ) uses a wa rehous e approach a nd i s an effec tiv e syste m fo r co llectin g, extracting, transform ing a nd cleaning organisatio nal data (C hiu 200 3). Th e ADC will facilitate the integratio n o f scien tific, fin ancial, hum an re source and c orporate datasets. T he EGIS will p rovide sp atial wareho use too ls to com plement the ADC. An Oracle relational datab ase is at the hub of th e AIMS data cen tre. Th e ab ility t o u se a commercial rela tio nal dat abase m anagem ent sy stem (DB MS) for a ll the data sto rage requ irem ents is p articu larly importan t in an en terp rise syste m (Zeiler 1 999 ). Th e DBMS is a repo sito ry for all data in th e syste m, in clu ding m etad ata. Set up in th is way, there is less overhead inv olved in m ainten ance and upd atin g o f AI MS datasets, du e to th e centralised s tora ge. Any changes t o the un derlying dat aset s are do ne once , in the D BMS, and t he up dat ed dat a is im medi ately avai lable t o all clients. Backups of the data are likewise sim plified. T he GIS s oftwa re us ed i ntegrates extrem ely closely with the DBMS, and is a ble to automatical ly u pdate sp atial metad ata as th e spatial d ata chan ges. Mu ltiv ersion ing of datasets is also possibl e, where diffe rent use rs can access diffe rent ve rsi ons of the sa me dat a. Thi s can b e easily i mplemented using th e curren t syste m, alth ough th is has no t been n ecessary at th is stag e. To pu blish th e resu lts of th e scien tific m odellin g in a m anner t hat satisfies the viewers int erests requires de dicated s oftwa re tools . At AIMS sev eral d edicat ed serv ers were b uilt ex pressly fo r this pu rpo se. Th eir con figuratio n is qu ite com plex with m ultip le ap plicatio n serv ers in teract ing as requests are received from viewe rs (fi gure 3). Initially th e re quest is in terp reted by th e Ap ach e web s erver. I f the sc ript cont ains C oldfusi on ele ments th en the Co ldfusion serv er in terp rets th e code otherwi se Tom cat Serv let Engi ne i nterpret s the co de. Coldfusi on by M acrom edi a Pt y Lt d i s a po we rful i nternet ser ver t hat pr ovi des rapi d depl oy ment of i nteract ive we b si tes. B oth ColdFusion a nd T omcat can send requ ests t o th e Arc IMS appl ication se rve r for speci fic ge ospat ial inform ation su ch as m ap gra phics a nd dat a array s. Figure 3. Architecture of the AIMS s patially enable d web si tes Th is web site can o perate with ou t Co ldfu sion instead usi ng HTM L an d Ja vascri pts t o provi de fun ction ality. Th e in tern al AIMS web site an d the external Reef Futures websit e contain m aps that are co nst ruct ed i n this m anne r, howe ver t he loadi ng t ime for t he Ja vascri pts and HTM L code is con siderab le and the ab ility to in teg rate with other pages is limited. C oldfusion se rve rs offer a fast a nd ra pidly cust om able interface that is engineered to interact with datab ases and mu ltip le scr ipting languag es. Th e page con tent is st ored in a SQL Ser ver 20 00 (M icrosoft) dat abase and i s com posed dy nam icall y based on the user ’s requests. For the we bm aps the Col dfusion code has b een eng ineered to prov ide sim ple spatial tool s s o the viewer can na vigat e an d i nterrogat e the inform ation on di spl ay (figure 4 ). Of particu lar no te is th e Ver ity search e ngine that is able to search an d index m ultip le do cuments to create a c ollection. Text searches ca n include t he entire document and no t ju st the m etad ata. www view er Apac he We b Server CF S erver To mcat Servlet Eng ine ArcIMS Application Serv er mon itor tasker ArcIMS Spat ial Serv er Spatial Dat a st ored in SD E Figu re 4. Sc reen ca pture of Web pa ges. The t op capture shows the regular inte rface with a toolbox and n avigation p ath on the left. Th e cen tral tex t contains buttons that provide access to a dditional inform ation such a s ot her web sites, p op up inform ation p anel s and we bm aps. The w ebm ap bel ow s hows the si mple tool s an d m ap l ayer managem ent wi th links t o addi tional m etadat a inform atio n. The application se rve r can delive r images, feat ure s, q uery and m etadata reque sts t o the spatial serv er. Th e spatial serv er then directly accesses the data to satisfy the re que sts. This ens ures re que sts are always ope rating on the curren t data. Ex tend ed fun ctio nality can be b uilt into th e viewi ng app licatio ns th rou gh the use of the Arc Ex ten sible Mark up Langu age (XML). XML is a flex ible con sisten t serv er-sid e lang uage that focuse s on inform ation transfe r. ArcXM L can be sen t directl y to th e app licatio n serv er with a soph isticated set of instru ction s larg ely det erm ined by the vi ewer. F uture de vel opm ent s in prov iding th e v iewer th e ab ility to manipulate th e modellin g in pu ts (ie th e te mperat ure of oc eans in the year 20 50 ) will b e thro ugh th is m ech anism . 2.2 User Fee dback With a web site th at is co mposed of m any activ e ele ments th e issu e of cognitiv e o verh ead and disorient ation dem ands at tention (B oechl er 20 01). Cognitive overhead is defi ned as the “am ount of cognitive resources neces sary to succe ssfully complete an in form atio nal task in hypertex t” (B oec hler 20 01, p. 27 ). Web sites that hind er the viewers ab ilit y to p lan ro utes th rou gh the web page , assi st wi th un derst anding t he c ont ents and coord inate in form atio nal task s will sig nifican tly degra de thei r e ffective ness. Hyp ertex t d isorien tatio n is the feeling of b eing lost with in th e stru cture o f t he web site. Sy mptoms of this d isorien tati on includ e lo op ing , ineffi cient navi gat ion, ge nerat ion o f que ry failu res, d isorgan ised screen layo uts with mu ltip le concurrent wind ow s, an d ex cessiv e back-tracking (Bo echler 20 01 ). To avo id cog nitiv e ov erh ead and d isorien tatio n the Reef Fu tures web sites will have cl ear navigat ional pat hway s wi th m inim al ‘clu tter’. Th e fi nal web site p roduct satisfies th e in itia l requ ests we receiv ed fro m le ading scien tists an d man agement au thorities. Th e Co llab orativ e Cen tre fo r Research on the Reef (CRC Reef) was able to en sure stak eho lders prov ided inpu t to th e websi te desi gn. T he si ngular m ost com mon request wa s to packa ge t he a vailable information into a sin gle site with s oph isticated search facilities. Fig ure 5 sh ows a simplified versio n of a suggested st ruct ure with c lear dem arcation of funct ion. Gi ven t he ca paci ty of hy pert ext we we re able to ob scu re the b ound aries b etw een the funct ions s o that al l fo ur options ca n be acc essed at any p oint with in the web site. On ce a to pic is selected vie wers are shown a front pa ge with an introdu cto ry st ate ment wh ich can be read lik e a brochu re ( opt ion 1, fi gu re 5 ). Al so included are the facilities t o search th e metad ata an d repo rts and t hen direct ly read those d ocum ents (o ption 2 & 3 , fi gu re 5). Fr om wi thin the broch ure page s and from the side m enu ‘toolbox ’ viewe rs can requ est th e in teractiv e m aps (o ption 4, figure 5). Figu re 5 . A s implified versi on of a re quest for web site fun ction ality. 3. CONCLUSION AND RECO MMENDATIO N Th e cap acity to delve in to the scien tific world is great ly en hanc ed t hrough t he use of t he AIM S kn owledge m anagem ent sy stem . The vi sual appeal that greets the vi ewer as they na vigate through th e mu lti-tiered Web Pages en hances th e overall effectiv eness. Th e si mplistic n ature of th e soft ware t ools ens ures t he wi der publ ic has complete acc ess with m inimal train ing. All elem ent s of t his KM S are gr owing rapi dly at presen t and th e fin al con figuratio n will b e a sin gle system wi th a com prehe nsi ve array of funct ions . Th rou gh this i nterface scien tists an d th e gen eral public will b e ab le to m aximize th eir invest igations ove r a wi de ra nge of topi cs. Wh ile th e syste m wo rks well, it is d ependent upon a nu mb er of co mplex in ter acting su b- system s, includi ng web serve rs, Uni x and Wind ow s Op erating Syste ms, h igh -sp eed netwo rks, GIS so ftware and th e DBMS. Initiall y there is a steep learning curve , and good com muni cat ion i s re qui red between t he Ne twork, GIS an d DB MS ad ministrato rs, particu larly as diffe rent pe ople fr om di ffere nt de part ments oft en perform th ese ro les. M aintain ing software compatib ilit y as co mpon ents are up grad ed is a continual area of frustr ation. For exam ple recent chang es fro m Co ldfusion 5 to C oldfusion M X have req uired installation o f an upgraded ArcIM S 4.01 . H owe ver t he A pache1 .3 web serve r software d id no t fu lly supp ort th is up grade and addition al co nfig uratio n was req uired. There are some aspects to the existing system that could be improved. Currently the software being used imposes a publishing process on the metadata before it can be searched using the ArcIMS Metadata Explorer Service. The original metadata is integrated with the datasets, and is automatically updated as the datasets change (eg.

spatial extent, projection, etc). To enable web- searches, the publishing pr ocess stores a copy of the metadata in the DBMS. This duplication of metadata is less than ideal, as metadata is not searchable in a web-based format until it has been published, and the published metadata must be re- published when the original datasets change.

The benefits for the scientific modelling community are substantial as integration of datasets mirrors the team-based research methodology. This research cohesion can generate substantial growth in the flow of knowledge both within the institute and to the wider community. Management agencies that wish to engage with scientists can begin to appreciate the complexities of the issues while scientists can ensure alignment with management priorities through feedback mechanisms. The wider community can begin to comprehend the reasons for scientific debate and uncertainty without the confusion surrounding program management and funding priorities. The influence of science should then be significantly enhanced as local management priorities are addressed within a regional environmental framework. Future research should address the ability to engage viewers with interactive modelling tools and multi-media presentations such as video interviews of key scientists. In situ web cameras could also be strategically employed (ie underwater at a bleached site) to add a temporal dimension to the KMS. Examination of the web server log files will provide rapid feedback on the tools and pages that viewers utilised. Web site development priorities should be based on these examinations combined with feedback comments.

The impact that the WWW will have on scientific KMS is only just being acknowledged and the potential to expand its functionality is immeasurable.

5. ACKNOWLEDGEMENTS The authors would like to thank the IT and Data Centre team lead by Scott Bainbridge for technical support; Terry Done, Vicki Harriott and Louise Goggin for direction; and two anonymous reviewers for constructive comments. 6. REFERENCES Boechler, P. M. How Spatial Is Hyperspace?

Interacting with Hypertext Documents:

Cognitive Processes and Concepts. Cyber Psychology & Behaviour 4 :23-46. 2001 Fink, P. Improving knowledge management systems:

issues and reviews of recent experiences.

OECD. 2002 Fischer, M. M. Innovation, knowledge creation and systems of innovation. The Annals of Regional Science 35:199-216. 2001 Freyfogle, E. T., and J. L. Newton. Putting science in its place. Conservation Biology 16 :863-873.

2002 Hlupic, V., A. Pouloudi, and G. Rzevski. Towards an Integrated Approach to Knowledge Management:‘Hard’, ‘Soft’ and ‘Abstract’ Issues. Knowledge and Process Management 9:90–102. 2002 Kininmonth, S. J. GIS:The ke y to research integration at the Australian Institute of Marine Science.

Pages 67-72 in J. Breman, editor. Marine Geography; GIS for the oceans and seas .

ESRI Press, Redlands. 2002 Koperski, K., J. Han, and J. Adhikary. Mining Knowledge in Geographic Data.

Communications of the ACM . in press Macromedia. Macromedia Coldfusion MX: Getting started building ColdFusion MX applications . Macromedia, San Francisco.

2002 Styhre, A. The Knowledge-i ntensive Company and the Economy of Sharing: Rethinking Utility and Knowledge Management. Knowledge and Process Management 9:228–236. 2002 Sweatman, H., A. Cheal, G. Coleman, S. Delean, B.

Fitzpatrick, I. Miller, R. Ninio, K. Osborne, C. Page, and A. Thompson.. Long-term monitoring of the Great Barrier Reef. 5, Australian Institute of Marine Science, Townsville. 2001 Talen, E. Bottom-up GIS. Journal of the American Planning Association 66:279-293. 2000 Williamson, A., and C. Liopoulos. The learning organisation information system (LOIS):

looking for the next generation. Info Systems Journal 11:23-41. 2001 Zeiler, M. Modelling our world . ESRI Press, Redlands, California. 1999