Related news calculating method and device based on fused keyword and theme model

The invention discloses a related news calculating method and device based on fused keywords and a theme model; the method comprises the following steps; selecting a first news URL and segmenting same; carrying out keywords extraction and theme model calculation according to segmented URL, thus obtaining keywords related information, a theme number and known news categories, and building a reverse index; selecting a candidate set according to the reverse index, ranking same, and selecting one or more second news related to the first news. The device comprises a segmenting unit, a calculating unit, an index building unit, a candidate set selecting unit and a ranking unit; the method and device can fuse news keywords with the theme model so as to calculate the related news candidate set, can employ different strategies to carry out rank and ctr feedback weight adjusting for different news categories, thus finally forming a whole set of related news calculating technical scheme. 本发明公开了种基于关键词和主题模型融合的相关新闻计算方法及装置，其