This chapter proposes a method first to return the implicit predicate and then to discover semantic relations between the implicit predicate and the two present nouns. For this purpose, a set of data resources is created and a programme is developed to automatically generate acceptable paraphrases for the Noun-Noun Compound (NNC). A web search goes through several major steps, including: accepting search words, sending search requests, filtering and sorting, searching indices, selecting web pages, ranking targets, and presenting results. The computer has been used in a number of studies to automatically interpret the attributive-head Chinese NNC and the key step is to restore the semantic relation which has been implied between the attributive and the head. According to different implicit predicates in different NNCs, different background events can be first recovered and then summarised into a library of paraphrase templates. The library is in turn used as resources with which the NNC can be automatically interpreted.