Data integration of several sources has gained considerable attentions with the recent popularity of the Web. In the real world, some information may be missing i.e., partial and some may be inconsistent from several sources. How to obtain information as complete as possible and detect inconsistency from these sources is thus an interesting question. Most existing work uses a simple graph-based or tree-based semistructured data model to represent heterogeneous data coming from various sites, which fail to account for the existence of partial and inconsistent information. In this paper, we rede ne the notion of semistructured objects to re ect the existence of partial and inconsistent information and study how to integrate such objects spread in various sources and check consistency in the meantime. We propose a new operator integration for this purpose and discuss its semantic properties.