2020年美赛C题O奖：Riddle of Sphinx: Cracking the Secret of Amazon’s Ratings and Reviews

【实例简介】2020年美赛C题O奖

Riddle of Sphinx: Cracking the Secret of Amazon’s
Ratings and Reviews

【实例截图】

【核心代码】

Contents
1 Introduction 3
2 Assumptions and Notations 4
2.1 Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Notations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3 Informative Evaluation Model 4
3.1 Vector Encoding Forms of Star Ratings . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.2 Contextual Entropy VADER Hybrid Model for Text-Based Measures . . . . . . . 5
3.2.1 Manually Annotating the Seed Word . . . . . . . . . . . . . . . . . . . . . 7
3.2.2 Contextual Entropy Block (CE) . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.2.3 VADER Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2.4 Proposed CE-VADER for Sentiment Analysis . . . . . . . . . . . . . . . . 9
3.3 Combination of Text-Based and Rating-Based Measures . . . . . . . . . . . . . . . 10
3.4 Model Implementation, Sensitivity Analysis and Results . . . . . . . . . . . . . . 11
4 Difference Equation to Measure Time-Based Pattern 11
4.1 Difference Equation Based Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4.2 Model Implementation, Sensitivity Analysis and Results . . . . . . . . . . . . . . 12
5 Predict Potential Success or Failure 14
5.1 Time Series Forecasting for Predicting Future Reputation . . . . . . . . . . . . . . 14
5.2 Evaluating the Success or Failure potential . . . . . . . . . . . . . . . . . . . . . . 14
5.3 Model Implementation and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
6 Specific Ratings and Descriptors Analysis 15
6.1 Specific Star Ratings Relevance to Rating Frequency . . . . . . . . . . . . . . . . . 16
6.2 Specific Quality Descriptors’ Relevance to Rating Levels . . . . . . . . . . . . . . 18
6.2.1 Naive Bayesian Model for Evaluation . . . . . . . . . . . . . . . . . . . . . 18
关注数学模型
获取更多资讯
2
MATHmodels
Team # 2002116 Page 2 of 38
6.2.2 Model Implementation and Results . . . . . . . . . . . . . . . . . . . . . . 19
7 Attractiveness Analysis of Design Features 20
8 Sales Strategies and Recommendations 20
9 Strengths and Weaknesses 21
9.1 Strengths . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
9.2 Weaknesses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
10 Conclusion 22
11 A Letter to the Marketing Director of Sunshine Company 22
Appendices 24
Appendix A Annotated Seed Words and Frequency 24
Appendix B The Number of Keyword Occurrences in Different Keyword Groups 25
Appendix C Top 1% Most Informative Ratings and Reviews 26
Appendix D Source Code for VADER Sentiment Analysis 36
Appendix E Source Code for Informative Algorithm 36
Appendix F Source Code for Reputation Calculation 37
Appendix G Source Code for Beyes Model 37
Appendix H Source Code for Time Series Prediction 38
Appendix I Source Code for Wordcloud Picture 38