[Purpose/significance] Text representation is an important step in intelligence processing. An excellent text representation model can reflect the document content precisely and sufficiently. Besides, it can improve the processing effect. It can be broadly applied in the fields of automatic abstracting and text segmentation. [Method/process] In this article, we collected the related documents and analyzed them. The construction methods and disambiguation in the lexical chain computing were classified and concluded. The computing method of the lexical chain relation included the computing method based on semantic association, the computing method based on statistical information and the computing method based on charts. The semantic disambiguation was important in the construction of the lexical chain, which directly affected the results and efficiency of the lexical chain construction. [Result/conclusion] The lexical chain text representation can be easily constructed and broadly applied. There are still some problems in the text representation model of the lexical chain. For example, there are many limitations to construct it by dictionaries, which does not take the context into consideration. The lexical chain model will possibly develop towards the fusion semantic relation method, the statistical algorithm and the context analysis of distributed semantics in the future.