Classification: You are so important!
Classification (multiclass, multilabel, binary, imbalanced…) might be the most widely-used computing technique in computational social science problems, e.g., social media post classification and user auto-labeling. We can see two lines of approaches. One is the rule-based and machine learning method, e.g., kNN, Decision Trees, and Naive Bayes. The other is the deep network-based approach. There are many fancy deep network models like bert, GPT, transformer, and attention model. I find three excellent articles:
I will give a tutorial at MSU Communication in Fall 2021, discussing all text classification techniques (hopefully) and text-based online trace data scraping/processing in computational social science from a practical perspective. Hope this talk will benefit communication researchers and myself because it definitely motivates me to recall classification algorithms, especially technical details, thoroughly and holistically. Codes and slides will be released soon.
Update: check out the tutorial