|
|
XISS Source Code for the VLDB'2001 paper: Click Here to Download XISS/R Source Code for the VLDB'2003 demo paper: Click Here to Download ================================================== Installation: of XISS ================================================== I have compiled the program on Linux and SunOS 1) You need to install the GIST library. You can get libgist from here: http://gist.cs.berkeley.edu/software.html Set the LIBGISTHOME variable to the correct path. 2) The parser uses libxml, so you need install it. You can get it from http://xmlsoft.org/ 3) Since the libxml may need zlib, you need it too. zlib can be found at: http://www.gzip.org/zlib/ 4) Change the path in the Makefile and Makefile.inc XIHOME should be set to the current directory 5) Change the include file path and additional library path in one of the following make files depending on your platform: Makefile.Linux.i686 Makefile.SunOS.i86pc Makefile.SunOS.sun4u Then you can try to make the binary files. Usage: 1) load xml files The program 'load' will load the xml files in the index directory. So, keep a directory named 'index'. They are some small xml files in data directory. You can have a try. 2) Query Join based algorithms: eejoin and eajoin They take five arguments: The first argument is the direct IO flag (0 or 1), if on SunOS platform it will use the direct IO feature. The second argument is the number of buffer pages. The third and fourth are the element names (or element attribute names) to be joined. The fifth is the level difference between the two elements in the xml document tree. Tree traversal based algorithms: They take the same arguments as join based algorithms. backtrace: bottom-up tree traversal element/element join eabacktrace: bottom-up tree traversal element/attribute join eatrace: Top-down tree traversal element/attribute join ================================================== Installation: of XISS/R ================================================== The loader was developed in C language using LibXML. Web interface and query processor were written in PHP. We also used Apache web server and the MySQL database. In order to run the program, you need to install a web server with PHP and MySQL support, and a MySQL database server. Copyright ©2000 The University of Arizona All Rights Reserved |
|