Study On Recommendation Technology Of Geocomputation Application Resources

Accompanied by the rapid development of high-performancegeocomputation,geocomputation application as an application mode of geocomputationprogram and data is created. The spatial data,analysis program,and application flowincluded in geocomputation application will be a sharp growth in the number andvariety,but our users really need is very limited.Therefore, to ease the geocomputationapplications resource overload problem,how to recommend resource which meet theuser’s need have become a key problem to be solved.In response to these issues,through researching in personalized recommendersystem, this article analyzed the pros and cons of content-based recommender algorithmand collaborative filtering algorithm; and according to the specialty of geocomputationapplication resources, geographical attributes and social labels are taken as importantreferences of personalized recommendation. The main work of this paper include thefollowing aspects:1. According to the characteristics of geocomputaton application resources,andapplication background of personalized recommender system,this work designed ageocomputation application resource model based on object-oriented designpattern.This model achieves the effective organization and management ofgeocomputation application resources,and lays foundation for the recommendation.2.After analyzing the geocomputation application resources which the user haveused,we found that geographic attributes(geotags) and social tags are connected withuser’s preference. Therefore,this work combined user’s preference of geographicalattribute and semantic similarities of his social tags, and implemented content-basedgeocomputation application resource recommender algorithm.And in view ofrecommendation need in the process of building geocomputaton application flow,afteranalysing the context information,the author implemented content-based andcontext-aware recommender algorithm.The experiment proved the effectiveness of thealgorithm.3.Based on the basic idea of collaborative filtering method, this work combineduser’s preference of geographical attribute, the quality and semantic similarity of socialtags, to get the target user’s nearest neighbor;recommended application resource withgeographic attribute similarity to the target user preference, and the probability that thetarget user attach a tag to it. The experiment proved the algorithm’s good effectivenesss.4.Based on the above research results,the personalized recommender algorithm ofgeocomputation application resources is integrated into the platform ofhigh-performance geographic computing as a plugin, and preliminarily meet the needsof the user obtaining geocomputation application resources.