To overcome challenges of processing timeliness and changeful topics when analyzing opinions of online news comments, free-tagging methods including two steps, object identification and polarity analysis, are proposed in this paper. In the first step, from news text, we employ the Affinity Propagation Cluster (APC) to extract key sentences whose name entities will be identified and their specified types will be identified as comment objects. After identifying comment objects, the method of distance-weight count of polarized words and semantic rules are used to determine the polarity of comments. Our method does not require labeled corpus, therefore it can reduce artificial workloads and avoid problems of non-versatility caused by changeful topics of comments. Experiment results show that our method is capable of identifying objects and polarity of online news comments with higher precision and wide coverage rate, and can satisfy the requirements of online news comments analysis.