Abstract: In the social network, the friendship is a kind of key element. With the changing of friendships, the social network evolves over time. And the information propagates widely with the friends interacting. Based on the personal properties and the status in social network, various behaviors are done by users, and the friendships are developed at the same time. As the frequent interaction is one the key character of friendship, it is proposed that the friends interaction behaviors are similar to each other, and the interaction mechanism is of reciprocity. Based on the time series analysis of user behaviors, interaction behaviors similarity computing model is proposed. Using the blog history snapshot data, the publishing, sharing, commenting and wall posting behaviors are evaluated, and then the similarity computing models are tested.

Key words: network behavior, interaction, similarity, friendship, blog

摘要： 朋友关系是构成社会网络的关键要素之一。朋友关系的不断变化，推动网络的发展。同时信息在朋友互动中得以广泛传播。用户依托其个体特质及其所在网络关系，开展丰富的网络行为，通过彼此互动，促进朋友关系的发展。鉴于频繁互动是朋友关系的重要特征之一，提出朋友在线互动行为之间具有较高的相似性假设，并遵循互惠机制。从行为的时间特性分析入手，提出用户互动行为相似性计算模型。利用博客的历史数据，分析发表、共享、评论、留言等互动行为，并对互动行为相似度模型进行验证。

关键词: 网络行为, 互动, 相似度, 朋友关系, 博客