Faced with huge amounts of information to realize the accurate retrieval under the network environment, the first step is indexing words cannot appear ambiguity word. Because Chinese’s the basic unit is Chinese characters, Chinese characters form words, Word is divided into monosyllabic word and compound word, and there’s no space between Chinese keywords and there are a lot of ambiguous concept. Therefore a lot of ambiguity in the indexing process will be produced. The result detected information of irrelevant or mistakenly identified. The paper focuses on a method to eliminating the crossed meanings ambiguous words in the automatic indexing. The paper puts forward a method to eliminating ambiguous words combined algorithm of exhaustive method and disambiguation rules. Experiments show that it can avoid a great lot segmenting ambiguities with better segmenting results.