Personal names are out of dictionary words which are usually primary keys in the web queries. Converting a personal name from source language to target language is transliteration. In this paper, we propose a novel algorithm for transliterating a Persian personal name to an English name. This method consists of two stages. At first, in the offline stage, a graph is made by processing Persian web contents. In this graph, names are related based on their neighboring in the web pages. Afterwards, in online stage, each name is transliterated using this graph and name frequencies. Experimental results show that the algorithm is effective in personal name transliteration.