Abstract: Inter-organizational commercial defamation often causes affected enterprises to suffer victim crisis against the backdrop of in-tensifying competition. Since commercial defamation is targeted, destructive and unclear in terms of responsibility, there is a perceived risk of purposefulness in impression management as to whether the strength of an enterprise's response is commensurate with the magni-tude of responsibility assumed. In this paper, we construct measures for crisis response tone analysis based on machine learning and sen-timent lexicon in the context of commercial defamation of A-share listed enterprises:action, certainty and tone intensity, and empirically test the effect and interaction of strong response tone of defamed enterprises in different audiences. The final experiment verifies the pur-poseful risk perception mechanism of “diffidence”. The results show that investors react negatively to the strong response of the defamed company after the business defamation incident, and the underlying mechanism is the perceived purposive risk of diffidence. Further re-search finds that journalists (vs. stock analysts) have a negative (vs. positive) attitude toward the tone of the strong response. Finally, different image dimensions have a moderating effect on the audience's risk perceptions. This study extends the study of defensive risk in-duced by crisis response tone to further deepen the understanding of corporate responses to different audiences.

Key words: competitive crisis, strong response tone, information filters, image dimension, investor reaction