| From the turn of the millennium,the idiom ’out with the old in the new’ could not be more relevant.This is the age of Information Technology which has seen rapid growth and expansion encompassing all regions of the globe.To date,the digital evolution can be broken down to three stages,each of them a profound leap over the last.First,the exponential advancement of computing hardware and power from which informatization was inevitable.Cue Moore’s Law,the Digital Universe arose at an historic moment.Second,the internet made possible a global system of interconnected computers and the spread of information.Suddenly relationships in the virtual world become as complicated and real as in the physical world.And on to now,the current period,which saw computer storage grow dramatically to the arrival of big data.Big data storage results in more efficient analysis and therefore offers greater benefits for the digital universe,stimulating it to brim over with vigor and vitality.From desktop to notebook,landline to smartphone,analogue television to the virtual reality headset,we are increasingly relying on the digital world,day in day out.There are few realms of the physical and social world which have not been digitalized.Data will become as fundamental a resource in the globalized economy as capital.However,while technology is developing and information is accumulating at a progressive rate,human cognition is finite and unable to utilize data completely.Information overload is become a growing concern.The Personalized Recommender System(PRS),spearheaded by the search engines,offers the user a tailored experience thereby circumventing the information overload problem.Regarding the overload problem,the PRS is a widely accepted solution having one huge advantage.The computer no longer needs to wait for the user to type in the search function.Instead,it is trying to understand users’ needs in advance.However this technology is still in its infancy,specifically the cold start problem and data sparsity.Unless these pressing concerns are addressed the quality of the recommendations will not improve.This paper advocates an algorithm inspired by the user-centered design process to tackle the overload problem.By doing so,the algorithm offers a more accurate and personalized real-time results.This thesis is structured as follows:Chapter 1.This chapter discusses some of the background information and problem which inspired our research before introducing our innovative solution.Chapter 2.Here I will present a literature review of the PRS as it stands today.Three important theories in our discipline will be reviewed as well as the recent user recommendation controversy.Chapter 3.This chapter provides a detailed analysis on the finalism,form,essence,and structural hierarchy of the PRS.Chapter 4.Extracting user preference data from visual hot spots on personal computer screens with a laundry list of methods such as Double Centerline Noise Reduction Method,Webpage Divisional Method,and keywords extraction method from short sentences.Chapter 5.A research model critique based on visual hotspots to determine user preference shall be presented,the outcome is an integrated user preference model which is made up of user instant preference,user short preference and user long preference.Chapter 6.Recent research on customer feedback models used by online retailers illustrate that disingenuous reviews are a genuine hindrance to accurate PRS.These findings are incorporated and elaborated upon into a new research model.Chapter 7.A method for better Real-time PRS which adapts to individual human interaction.Chapter 8.Summary of my findings,and implications for the future.What is worth mentioning,this article is innovative in a bunch of ways,but on the whole it is divided into two levels:Theoretically,we have clarified that PRS is cognitive assistant and not an algorithm.We proposed the WHW model,which explains what to recommend,how to recommend and when to recommend it.This model also considered combining real-time,short term and long term interest together to figure out users’ preference.Methodologically,in the first instance,we determine virtual hotspots via psychological experiment,then extract users’ instant interest from their real-time browsing behavior with the benefit of Double Centerline Noise Reduction Method,Webpage Divisional Method and keywords extraction method.In the next place,an integrated user preference model comes into being which is made up of user instant preference,user short preference and user long preference with the assistance of cluster and session segmentation method.Once more,we construct online reviews’ refine solution based on Appraisal Theory and Discourse Markers in literature theories.Ultimately,we develop a prototype PRS and verify the capability,availabity and instantaneity of Interactive and Convergent Algorithm for Personalized Recommendation. |
PDF Full Text Request