A hybrid algorithm to track drift of user's interests
YI Gai-zhen · Caai Transactions on Intelligent Systems · 2010
User interests inevitably drift while using a personalized information service.A new forgetting function with normal distribution density was proposed to accommodate drift.The function conformed to a user interest forgetting law.User interest was defined in a hybrid model that contained both long and a short-term components.The short-term component was renewed by using the least recently used algorithm.The long-term component was renewed by using the normal incremental forgetting distribution algorithm.Experiments showed that the algorithm noted changes in user's interests more quickly and tracked them more accurately,greatly improving the efficiency of personalized information services.