Notice of RetractionThe application of improved K-Nearest Neighbor classification in topic tracking

News topic tracking is one of the major tasks of TDT, the aim of which is monitoring news story flow as well as recognizing and giving subsequent stories in advance related to topics described by several news stories. This paper explains in details the concept of topic tracking and the common methods at present; aiming at the scarcity of positive example of training, this paper makes effective improvement on traditional KNN taxonomy and applies it to topic tracking; besides, it adds time window strategy to the process of topic tracking, which effectively reduces calculation complexity. The final experimental result also proves that this method is superior to traditional KNN topic tracking method.