As a new economic development model to cope with the threat of resource shortage and the demand for green and low-carbon transformation, circular economy is receiving more and more attention. In order to promote the practical process of this plan, many measures have been proposed and put into action, and the construction of a waste material recycling system is one of them. Considering that the decision-making process related to such problems usually involves many constraints and needs to balance the opinions of all parties, this paper proposes a multi-attribute group decision-making method based on linguistic preference ordering and extended TOPSIS technique. Firstly, this study uses data mining and natural language processing technologies to crawl relevant news websites and obtain a large amount of public data, then determines the attributes and their weights through keyword extraction and clustering. Secondly, the multi-criteria mutual evaluation matrices are proposed and used for determining the weights of experts involved. Thirdly, this study employs the linguistic preference ordering method, an extended TOPSIS method and a min-max optimization model to successively achieve the complete decision-making process of decision information representation, individual ranking and collective ranking. Finally, this paper applies the proposed method to the example of setting schemes selection of waste material recycling site in order to verify its effectiveness and applicability, and further elaborates its advantages and characteristics through comparative analysis.