Role:
Frontend Implementation
Many TV series synopses only provide the beginning of the plot to attract viewers, but this often leads to audiences watching shows that end poorly. This project extracts key plot points from each episode and creates an emotional visualization interface for deeper analysis of the TV series. Its innovation lies in analyzing audience reviews to offer potential viewers a better understanding of the show, while also helping the promotion team adjust strategies based on emotional trends in the comments and providing investment references for investors.
bs4, re, and urllib are employed to parse web pages and extract review content. To counter the website's anti-scraping mechanisms, the project uses methods such as adding header information and setting proxy IPs to ensure stable and efficient data scraping. Ultimately, the scraped review data is stored in a MongoDB database, providing foundational data support for subsequent analysis.Data Analysis and Processing: In the data analysis phase, the project first uses the
jieba word segmentation tool to process the review text and employs the TF-IDF algorithm to extract keywords. At the same time, data cleaning is performed to remove invalid or low-quality comments. Next, a multi-layer bidirectional GRU neural network is constructed using the TensorFlow framework to perform sentiment analysis on the comments. The analysis classifies the sentiment into categories such as joy, anger, disgust, and sadness, enabling quantitative analysis of the sentiment tendencies of the comments. This provides core data support for visualization.Follow the principles of visual coding design and select appropriate chart types for the obtained data, including bar charts, line charts, rose charts, radar charts, and word clouds.
Through the episode discussion popularity line chart, the publicity team can monitor the changes in the TV drama's popularity in real time and adjust marketing strategies when popularity dips. At the same time, this chart helps investors predict the trend of popularity. The sentiment radar chart and sentiment trend bar chart analyze the emotional distribution and changes of the TV drama from both the overall and time dimensions. During the airing, these can be used for public opinion guidance, and after the airing, they can summarize the emotional patterns of similar TV dramas. Additionally, the dynamic word cloud reveals the hot topics that viewers are focusing on through high-frequency words, providing the production team with promotional direction and supporting the precise marketing of the TV drama.