Most of the social media such as Instagram/Facebook/Flicker/Twitter allow users to express themselves through the post content, like content and profile information. Each of these activities is a reflection of a user's personality. Images are currently present in a significant fraction of data in social media. Therefore, these days personality trait prediction by image-based content analysis attract more attention than text-based analysis among researchers. However, investigating of large amount of unstructured data of presented images is complicated. Therefore, proposing models that can prepare this knowledge briefly and structured is crucial. In this paper has been attempted to detect personality by combined deep learning and clustering approach that accurately inferring relationship between users’ images content and their personality in Flicker. We considered different modalities of interaction including: posts, likes and their profile images in this platform. The results of this paper will be useful in different areas such as marketing advertisement, psychology and sociology.
