[extensive reading of papers] analyzing connections between user attributes, images, and text

Thesis title ：Analyzing Connections Between User Attributes, Images, and Text
Time ：2020
source ：part of Springer Nature
There is no public code in this paper
There is no public code in this paper
There is no public code in this paper

I understand what it does is ： Images and text uploaded by users , Judge the user's personality and gender
I mainly want to find out from this article ： How to extract image attributes ？ How image attributes and text attributes are combined ？
Abstract
   This work explores personal information and psychological characteristics ( for example , Gender and personality ) And the relationship between self-identity image and words . We use N≈1350 Image and title data set , Automatically extract features from images and titles . We Several image and text attributes have been identified , It shows a reliable relationship between participants and individual differences . Besides , We consider the task of using unimodal and multimodal features to predict gender and personality . We prove that the multimodal prediction method is superior to the pure visualization method and the pure text method . We believe that , Users upload billions of pictures every day , Our research on the relationship between user characteristics and user data is related to online settings .

A version of Pearson's correlation coefficient is used to calculate the correlation between personality and visual and textual features .

You can continue to look at the references ：
48.Zhang D, Islam MM, Lu G. A review on automatic image annotation techniques. Pattern Recognition. 2012;45(1):346–362.
Overview of automatic image annotation technology