Every resource in social tagging system have hundreds even thousands tags. To every certain resource, tags always have different popularity. The higher popularity a tag has, the more it suit for represent features of resource. This paper first find Top30 tags are popular tags, but the average tagging rate is quite low while approaching the 30th tag. Thus, six groups of resource that have vary saved times are taken for further analyzing. In all six groups, Top9 tags have high ATR, as well as ATR deviation, which mean these tags have obvious advantages while compare with others. Thus we take Top9 as representative tags. Finally, a user profile construction algorithm is given out based on representative tags.