Nowadays, the Internet is growing rapidly and the number of network data is also growing rapidly, which makes it more difficult to get information from the massive network data with traditional methods. Entity relation extraction is an important research direction of natural language processing. It can find and identify the semantic relation of the entity to analyze the abstract textual data. Unstructured network data can also be transformed into structured data by using Entity relation extraction. This paper presents an unsupervised entity relation extraction model, which can overcome the shortcomings of traditional methods, such as needing a lot of man-made work and poor portability. In this model, we create a filter function at first. Then we extract the relational feature words by using context window and parsing. We use affinity propagation clustering algorithm to get the relation of entities, which can obtain better results than k-means clustering algorithm.