For detecting review spam in the Internet automatically, an method based on resistance network is proposed. By treating the distance between two reviews as a resistance, we represent the given dataset as a resistance network and the resistance distance between two nodes is a measure of the semantic distance between them. Spam reviews are semantic outliers in this network. An electrical outlier factor (EOF) for each review based on its influence on the power dissipated of the network was used to detect the spam reviews. Experimental results testified that EOF is suitable for detecting review spam, and is efficient and effective．