Abstractive multi-document summarization aims at generating new sentences whose elements originate from different source sentence. It can be achieved via phrase selection and merging approach which aims at constructing new sentences by exploring syntactic units such as fine-grained noun and verb phrase. It can be also achieved by extracting semantic information from source sentence which uses the concept of Basic Semantic Unit (BSU) and semantic link network. Clustered semantic graph approach employs semantic role labeling and predicate argument structure to construct the summary. These approaches aim at generating efficient abstractive multi-document summarization. This paper presents the merits and demerits of the above methods in the context of abstractive text summarization.