A User Acceptance of Web Personalization Systems

Research on web personalization  techniques for collecting and analysing web data in order to deliver personalized information to users is in an advanced state.  Many metrics from the computational intelligence field have been developed to evaluate the algorithmic  performance of Web Personalization Systems (WPSs).  However, measuring the success of a WPS in terms of user acceptance  is difficult until the WPS  is deployed in practice.  In summary,  many techniques  exist for delivering personalized information  to a user, but  a comprehensive  measure of the success in WPSs in terms of human  interaction and behaviour  does not exist.

This study  aims to develop a framework for measuring user acceptance  of WPSs from a user perspective.  The proposed framework is based on the unified theory of acceptance  and  use of technology  (UTAUT). The  antecedents  of user accep- tance  are described by indicators  based on four key constructs, i.e. performance expectancy  (PE),  effort expectancy  (EE),  social influence (SI),  and  facilitating conditions  (FC).  All these  constructs  are underpinned  by Information  Systems (IS) theories that  determine  the intention  to use (BI) and the actual  use (USE) of a technology.

A user acceptance  model was proposed and validated  using structural equation modelling (SEM)  via the  partial  least  squares  path  modelling (PLS-PM). Four user characteristics (i.e.  gender, age, skill and experience) have been chosen for testing the moderating effects of the four constructs. The relationship between the four constructs  in regard to BI and USE has been validated  through  moderating effects, in order to present an overall view of the extent of user acceptance  of a WPS.

Results from response data  analysis show that the acceptance  of a WPS is deter- mined through  PE, EE SI, and FC. The gender of a user was found to moderate

the relationship between performance expectancy of a WPS and their behavioural intention in using a WPS. The effect of behavioural  intention  on the use of WPS is higher for a group of females than  for males. Furthermore, the proposed model has been tested  and validated  for its explanation  power of the model and effect size. The current study  concluded that  predictive  relevance of intention  to use a WPS is more effective than the actual WPS usage, which indicated that  intention to use has more prediction  power for describing a user acceptance  of a WPS.

The  implications  of these measures  from the  computational intelligent point of view are useful when a WPS is implemented.  For example, the designer of a WPS should consider personalized design features that  enable the delivery of relevant information,  sharing to other users, and accessibility across many platforms, Such features  create  a better  web experience and  a complete  security  policy.  These measures can be utilized to obtain  a higher attention rate  and continued  use by a user; the features that  define user acceptance  of a WPS.