| | |

INFLUENCE OF RATIO OF AUXILIARY PAGES ON THE PRE-PROCESSING PHASE OF WEB USAGE MINING


Information management

INFLUENCE OF RATIO OF AUXILIARY PAGES ON THE PRE-PROCESSING PHASE OF WEB USAGE MINING

Name and surname of author:

Michal Munk, Ľubomír Benko, Mikuláš Gangur, Milan Turčáni

Year:
2015
Volume:
18
Issue:
3
Keywords:
Web usage mining, data pre-processing, session identification, auxiliary pages, reference length, log files, business intelligence, data mining
DOI (& full text):
Anotation:
Data mining belongs to the one of the important tools for Business Intelligence. It is a means to increase competitiveness of a company. Web usage mining is engaged in data mining of web server log file and it analyzes the user´s behavior on the web site. The first step of web usage mining process is data pre-processing obtained from a web log file. Data pre-processing is an important part of web usage mining. Discovering patterns of behavior of web visitors depends on the quality of pre-processing phase. Therefore it is important to understand the used methods. This paper summarizes the pre-processing phases and especially the phases of session identification. There are introduced two algorithms for data cleaning and session identification using the reference length method. The main aim of this paper is to compare a calculation of cutoff time and its influence on discovered useful, trivial and inexplicable rules. Cutoff time is an important part of the session identification using the Reference Length method. The influence of ratio of auxiliary pages on the calculation based on a sitemap and subjective estimation was compared. Statistical methods were used to determine the difference between these two approaches. In this paper was examined the portion of found rules based on quantity and quality. The ratio of auxiliary pages has only an impact on quantity of extracted rules in the files with path completion. It has no impact on portion of extracted useful rules, on the other hand, inappropriate estimation of the ratio of auxiliary pages may cause increasing of trivial and inexplicable rules.
Data mining belongs to the one of the important tools for Business Intelligence. It is a means to increase competitiveness of a company. Web usage mining is engaged in data mining of web server log file and it analyzes the user´s behavior on the web site. The first step of web usage mining process is data pre-processing obtained from a web log file. Data pre-processing is an important part of web usage mining. Discovering patterns of behavior of web visitors depends on the quality of pre-processing phase. Therefore it is important to understand the used methods. This paper summarizes the pre-processing phases and especially the phases of session identification. There are introduced two algorithms for data cleaning and session identification using the reference length method. The main aim of this paper is to compare a calculation of cutoff time and its influence on discovered useful, trivial and inexplicable rules. Cutoff time is an important part of the session identification using the Reference Length method. The influence of ratio of auxiliary pages on the calculation based on a sitemap and subjective estimation was compared. Statistical methods were used to determine the difference between these two approaches. In this paper was examined the portion of found rules based on quantity and quality. The ratio of auxiliary pages has only an impact on quantity of extracted rules in the files with path completion. It has no impact on portion of extracted useful rules, on the other hand, inappropriate estimation of the ratio of auxiliary pages may cause increasing of trivial and inexplicable rules.
Section:
Information management

?
NAPOVEDA
reguired