| Popis: |
With the emerging of big data, the sci-tech information research is of great significance. Internet moves information at ever-increasing speeds along a nearly infinite set of pathways, but also makes the acquisition of effective knowledge hard and time-consuming. To address the issues of information redundancy and knowledge omission, in the paper we design a pushing system of cutting-edge sci-tech information, which could quickly find the targeted knowledge with the same topic. We use cosine distance to compute the similarity between two information clusters. The cosine distance based indexes accelerate the determination of cutting-edge literatures and maintain the timeliness of information retrieval. Through our three-month operation and multidimensional evaluation, the proposed cosine distance method shows 21% better effective ratio than Euclidean distance and more than 3 times faster than enumeration method. |
