Evaluation of game contents using SNS data and Word2Vec

Ik-Joong Kim, Jung-Yoon Kim
2018 Journal of Next-generation Convergence Information Services Technology  
SNS is useful place for infer evaluation based on user generated feedback regarding various contents.However, SNS data contains many profanity and internet slangs, and wrong grammar. So it makes difficult to understanding and analyze meaning by the computer. In this paper, we collected SNS data for specific game users for 6 months. Then, Word2Vec was used to analyze the meaning of game contents to users. SNS data compares the distance between sentences including CBOW and Skip-gram, and suggests a reasonable machine learning model for SNS data.
doi:10.29056/jncist.2018.06.04 fatcat:fs4zdipn4zazxnba6nvhz5piim