However, it is important to note that the load balancing 
middlewares in the current solution check/scan SESQL 
sentences one by one. In order to provide a better performance, 
neighbouring SESQL sentences might be checked and 
considered together. In the meantime, we are aware that the 
decision rules identified are still quite simple and static, and 
needed to be improved further. Currently, we are trying to use 
machine learning methods to mine the application logs to 
improve the current rules. 
5. CONCLUSIONS AND FUTURE WORK 
Spatial analysis plays a key role in GIS. In order to meet the 
increasing demand of spatial information applications in the 
Web, spatial analysis should be provided into WebGIS. This 
article aims to incorporate load balancing spatial analysis into 
XML-based WebGIS. Compared to other approaches that 
implement spatial queries and analyses solely on the server or 
browser sides, load balancing spatial analysis carries out spatial 
analysis either on the server or the browser sides depending on 
the network communication cost and the computational cost. 
This article mainly focuses on designing the load balancing 
middlewares. Comparison with other solutions shows that the 
proposed load balancing solution can optimize the execution of 
spatial analysis, and therefore lead to a better performance. 
The contributions of this article are: 1) proposing a finer 
granularity (by spatial objects) of transmitting spatial data in 
load balancing middlewares; 2) identifying more flexible and 
precise decision rules for distributing spatial operations to 
server or browser sides; 3) enabling high-performance spatial 
analysis in XML-based WebGIS. 
Our next step is to evaluate the proposed solution with more 
complex case studies. We are also interested in using machine 
learning methods to identify decision rules from application 
logs. Query optimization in SQL research might also provide 
some hints on improving the load balancing middlewares. 
Furthermore, the proposed solution is not only suitable for 
XML-based WebGIS. Therefore, application of the proposed 
solution in other WebGIS will be investigated. 
REFERENCE 
Chang, Y. Park, H. 2006. XML Web Service-based 
development model for Internet GIS applications. 1JGIS, 20(4), 
pp. 371- 399. 
Huang, H., Li, Y., Gartner, G., Wang, Y., 2011a. An SVG- 
based method to support spatial analysis in XML/GML/SVG- 
based WebGIS. /JGIS, 25(10), pp. 1561-1574. 
Huang, H., Li, Y., Gartner, G., 201b. A load balancing method 
to support spatial analysis in XML/GML/SVG-based WebGIS. 
In: Advances in Web-based GIS, Mapping Services, and 
Applications, Li, S. Dragicevic, S., Veenendaal, B. (Eds.), CRC 
Press, pp. 153-168. 
Levine, J. Mason, T., Brown, D. 1992. Lex & Yacc (2nd). 
O'Reilly & Associates. 
Lin, H., Huang, B. 2001. SQL/SDA: A query language for 
supporting spatial data analysis and its web-based 
implementation. IEEE Transactions on Knowledge and Data 
Engineering (TKDE), 13(4), pp. 671-682. 
Peng, Z., 1997. An assessment of the development of internet 
gis. In: Proceedings of the ESRI User Conference. 
Peng, Z., Zhang, C., 2004. The roles of geography markup 
language (GML), scalable vector graphics (SVG), and Web 
feature service (WFS) specifications in the development of 
Internet geographic information systems (GIS). Journal of 
Geographical Systems, 6(2), pp. 95-116. 
SuperMap, 2010. SuperMap IS.NET 2008. http://www 
.supermap.com.cn/gb/products/fwskf.htm (Mar. 2010). 
Vatsavai, R., Shekhar, S., Burk, T., Lime, S., 2006. UMN- 
MapServer: A High-Performance, Interoperable, and Open 
Source Web Mapping and Geo-Spatial Analysis System. In: 
GIScience 2006, Raubal, M. (Eds.), Springer, pp. 400-417. 
Wu, X. 2002. Principles and methods of GIS (in Chinese), 
Publishing House of Electronics Industry, Beijing. 
APPENDIX. SESQL SENTENCES IN CASE STUDIES 
Case Study 1 
1) Calculate a 20 km buffer of river “R1”: 
create view bf as select id, buffer(d, 20000) as buf from river 
where ID="R1" 
2) Find out all the districts which overlay with this buffer : 
create view citylist as select boundary.id as id, boundary.d as d 
from boundary, bf where (overlap(boundary.d, bf.buf) true) 
3) Find out all the affected areas in each district: 
Create view affectedarea as select citylist.id as id, intersection 
(citylist.d, bf.buf) as d from citylist, bf 
4) Calculate the size of affected area in each district: 
Select id, area(d) as area from affectedarea 
Case Study 2 
1) Compute the railway segments in each district: 
Create view seg r as select boundary.id as id, intersection 
(boundary.d, railway.d) as int from boundary, railway 
2) Calculate the length of each railway segment: 
create view seg r. len as select id, length(int) as len from seg Y 
3) Sum up all the railway segments for cach district: 
create view seg r total as select id, sum(len) as total len from 
seg r len group by id 
4) Filter out districts which have no railways: 
Select id, total len from seg r total where total len>0 
Case Study 3 
1) Calculate a 20 km buffer of railway “Guang-Mei-Shan”: 
create view buf as select id, buffer(d, 20000) as buf from 
railway where (id='Guang-Mei-Shan') 
2) Identify the districts whose centres are located in this buffer: 
Select city.id from city, buf where (within(city.d,buf. buf) -true) 
     
  
    
  
  
  
  
  
  
     
   
  
  
  
  
   
  
  
  
  
   
   
   
   
  
   
     
  
    
   
    
  
    
   
   
  
   
    
  
   
   
   
  
  
    
   
   
    
   
    
    
      
Dept. of 
KEY WC 
ABSTRA 
World-wi 
unobserve 
disaster m 
model bec 
this issue, 
publish/su 
However, 
services ai 
data with 
based on 
sensor wel 
time senso 
l1 Back 
The Worlc 
tremendou 
from vide 
matchbox- 
environme 
scientists 
unobserval 
attracting | 
habitats 1 
environme 
structure | 
application 
(Kassab er 
them are ti 
However, 
traditional 
inefficient 
between us 
publish/sul 
1987) pro: 
Queries anc 
notification 
While the | 
Other disci 
New in the 
this slow a 
are current] 
— 
* 
Correspi