Along with the rapidly development of the information retrieval and web technology, web entity retrieval has become a new popular way for getting specific information, such as looking for a book or a movie. Like document retrieval, generally there are too many results returned for a query, so ranking is still a necessary step during the entity retrieval process. This paper will focus on the ranking problem for web entity. Two methods are proposed, the first one will rank the results by a relevance score directly and the second one get the final ranking list by a training model. To compare the effective of the two methods, by the same features, we perform related experiments. According to the test data from real web pages, we test the precise of each method and get the conclusion at last.