关键词: 企业环境行为, 对偶犹豫模糊集, 对称交互熵, 相似度, 聚类

Abstract: Accompanied by increasingly serious ecological and environmental problems, the whole society's awareness of ecological protection is gradually strengthened. Enterprises, as the main source of pollution, are paid more attention to in the market, and thus all kinds of corporate environmental behavior indicators as the information to understand the environmental protection efforts of the enterprise, are gradually included in investor considerations. This paper focuses on the third-party international environmental auditing platform to evaluate the environmental behavior of enterprises, researches the clustering method of corporate environmental behavior information for dual hesitant fuzzy evaluation information, and identifies and selects excellent enterprises in this regard. First of all, this paper combs through the relevant research literature on enterprise environmental behavior at home and abroad in recent years, and agrees that the decision-making of enterprise environmental behavior belongs to the multi-attribute problem, and determines that the fuzzy mathematical theory will be used next to conduct research on enterprise environmental evaluation.
Firstly, based on the third party international environmental audit process, this paper proposes that the dual hesitant fuzzy set consisting of two sets of possible values of affiliation and non-affiliation has a relative advantage in solving this kind of multi-attribute decision-making problem. Referring to related literature, based on the definition of dual hesitant fuzzy distance measure, we construct dual hesitant fuzzy relative entropy and dual hesitant fuzzy symmetric interaction entropy, and prove that the symmetric interaction entropy has non-negativity, symmetry, and uniqueness of the value of zero. Then, a new dual hesitant fuzzy similarity formula is proposed based on the information theoretic perspective, and its five properties suitable for use as an evaluation criterion are clarified, and it is further extended to a similarity formula considering weights. Finally, the similarity coefficient matrix is constructed, and with reference to the braided net clustering method for hesitation fuzzy sets, a braided net clustering step for dual hesitation fuzzy sets is proposed to cluster the information in the dual hesitation fuzzy sets.
On the basis of the references, this paper makes appropriate modifications to the cases therein, takes into account the case of enterprise environmental assessment by the third-party international environmental audit platform formulated by the government, and evaluates the degree of environmental behavior of enterprises under the weights of different indexes as an example of calculation, evaluates and screens enterprises with high environmental return on investment ratio, which proves the correctness and feasibility of the method in this paper. Afterwards, by comparing the results with those obtained by using the traditional correlation coefficient clustering method, it shows that in the consideration of the large amount of information provided by decision makers, the method in this paper is clear in concept, simple in calculation, and high in resolution, which relatively makes up for the shortcomings of the use of dual hesitation fuzzy correlation coefficients in the references, and at the same time takes into consideration the weights of the evaluation indexes of the enterprise's environmental behavior, which is more in tune with the evaluation of the market's environmental behavior in the realistic scenario. In general, the dual hesitant fuzzy enterprise environmental behavior decision-making model considering symmetric interaction entropy is a more flexible and comprehensive multilevel evaluation and selection model, which provides a new way of thinking to improve the construction of enterprise environmental evaluation system.
In the future follow-up research, the correlation coefficient and the distance definition of dual hesitant fuzzy set in statistics can be further combined to improve the similarity of the correlation coefficient in the statistical sense, and the real environmental behavior data of enterprises can be collected for empirical testing, so that the dual hesitant fuzzy set can be further developed more fully in integrating the environmental decision-making, evaluation and selection of enterprises.

Key words: corporate environmental behavior, dual hesitant fuzzy sets, symmetric cross entropy, similarity, clustering