Web Utilization Miner WUM 7.0 Beta (Demo) ========================================= Information ----------- WUM home page: http://wum.wiwi.hu-berlin.de WUM download: http://wum.wiwi.hu-berlin.de/download WUM online demo: http://wum.wiwi.hu-berlin.de/demo WUM bug submission: http://wum.wiwi.hu-berlin.de/bugs WUM tutorial: http://wum.wiwi.hu-berlin.de/manual (Coming soon!) 1. Do you have any questions concerning the installation of WUM? Did you get any kind of Java exception during the execution of WUM? E-mail Karsten Winkler at kwinkler@ebusiness.hhl.de! 2. Do you have any questions on how to apply WUM on your web log? Do you need help in deciding how to prepare your file for WUM or in forming mining queries? E-mail Myra Spiliopoulou at myra@ebusiness.hhl.de! Quick Installation Guide ------------------------ (0) Software requirements: The Java Development Kit 1.2.2 (Java 2) or higher must be installed. (1) Choose the compressed file to download: (Unix or Linux) WUM.v70.tar.gz or (Windows 95/98/NT/2000) WUM.v70.zip Otherwise, download the file that is specified in the WUM registration mail. Make sure to type the exact file name or use cut&paste instead to access the given download URL. (2) Create a directory for the Web Utilization Miner WUM and copy the appropriate compressed file in that directory. Make sure that the new directory is now your current working directory. The user starting WUM must have write permission in this directory. (3) (Unix or Linux) Unzip and untar the compressed file: % gunzip WUM.v70.tar.gz % tar -xf WUM.v70.tar (Windows 95/98/NT/2000) Unzip the compressed file in your working directory using e.g. the shareware WinZip (http://www.winzip.com). (4) Modify the environment variables JAVA_HOME and WUM_HOME in either (Unix or Linux) wumgui and wumbatch or (Windows 95/98/NT/2000) wumgui.bat and wumbatch.bat according to your needs. (5) (Unix or Linux) Make ${WUM_HOME}/bin your current working directory and start the Web Utilization Miner using the script: % ./wumgui (Windows 95/98/NT) Start the Web Utilization Miner by running the batch file wumgui.bat that is located in your working directory. This batch file will set the environment variables. You might have to create a PIF file wumgui.pif in order to increase the environment memory associated with the batch file. Please ask your local system administrator how to create this file. The wumgui.pif file can be created by starting the Windows Explorer and changing the properties of the file wumgui.bat. Double-click wumgui.pif in order to start the miner WUM. (6) Choose 'File > Create Mining Base' to create the small demo mining base: Name: Demo Web Server: http://www.demo.org Directory: ${WUM_HOME}/data/demoWebSite/ Log Files: ${WUM_HOME}/data/demoWebSite/ The term ${WUM_HOME} must be replaced by the appropriate directory name. You might click the button '...' to choose a directory using the file dialog. Click 'Create' to create the new mining base in WUM. A new Object Store PSE database and various auxiliary files will be created in the mining base directory. The directory containing the log files can differ from the mining base directory. At the moment, log files cannot be accessed via FTP. Choose 'File > Close Mining Base' to close the current mining base. (7) Choose 'File > Open Mining Base', select 'Demo' and click 'Open' to open the corresponding mining base in WUM. The directory ${WUM_HOME} contains a small generic log file for demostration purposes. The structure of the example Web site ist depicted at the folling URL: http://wum.wiwi.hu-berlin.de/demo/ (8) Choose 'File > Import Log File', specify the log file name either ${WUM_HOME}/data/demoWebSite/AccessLog.txt and the log file format (x) Common Format for the common log file format and click 'Import' to import the demo access log file. Once again, the term ${WUM_HOME} must of course be replaced by the appropriate directory name. Do not cancel the import process. A short summary will be shown after the log file is successfully imported into the WUM mining base. (9) Choose 'Analyzer > Create Visitors' Sessions' and click 'OK' to determine the users' 22 sessions that are contained in the demo log file. Please use the proposed time threshold for the maximal session duration (30 minutes: '0/00:30:00'). (10) Choose 'Analyzer > Create Aggregated Log (Database)' and click 'OK' to generate the aggregated log. The default setting 'Exclude Referrer URL in Observations' should not be changed. All users' sessions will be aggregated in this data structure. The aggregated log consists of 18 observations and 3 children of the root node and is persistently stored in an Object Store PSE Pro database. (11) The 'Visualizer' can now be used to view single sessions, observations (i.e., branches of the aggregated log) and the complete aggregated log as a tree. (12) Choose 'Queries > MINT Ad Hoc Processor (Database)' to query the aggregated log using the mining language MINT. The mining language is described at the WUM home page. Make (currently) sure that all elements of the MINT syntax are typed in small letters. All parentheses, operators, variables etc. must be separated from each other using blank spaces. Choose 'Help > MINT Syntax' to view a short summary of the implemented MINT Syntax and a few example queries that are not related to the imported log file. After entering the MINT query in the text field 'MINT Query', it can be executed by clicking the button 'Execute'. Use the button 'Pages' to to view all Pages of the example Web site that are contained in the database. The attribute access of each page denotes the total number of accesses by all visitors in the log file. Processed queries can be saved and later be restored using the buttons 'Save' and 'Load'. (13) If your log is too big for WUM, there are solutions: 1. Our (UNIX only) filtering preprocessor. 2. We can give suggestions on what to filter out and how to instruct the preprocessor to do so. 3. We can give suggestions on how to split the log without blurring its statistic properties The preprocessor is not part of the package. E-mail Myra Spiliopoulou at myra@ebusiness.hhl.de Please remember, the Web Utilization Miner is still a Beta-Version! The WUM team would appreciate any bug submissions or feature suggestions concerning the future development of the miner. Please visit the following Web page: http://wum.wiwi.hu-berlin.de/bugs/. Thank you in advance! Kind regards, Karsten Winkler