Please sign up or log in to continue.

Analyzing stock market movements based on sentiment analysis

Utilizing NLP for sentiment analysis of stock comments, and analyzing the sentiment of comments enables the assessment of whether the stock is viewed bullish or bearish by the score of the comments.

Comment

Inspiration

By a stroke of luck, one of my teammate stumbled upon a tech website and discovered a small application called Sentiment Analysis. I gave it a try and found that the sentiment analysis wasn't accurate. Later, I reached out to the developer of the application, and after some debugging, the results improved. Since our company has some involvement with bonds, we had the idea of applying sentiment analysis to stock comments and analyzing the correlation between the sentiment scores of stocks and their bullish or bearish outlook. And thus, we embarked on this Hackathon journey.

What it does

Importing a dataset of stock comments, performing sentiment analysis on each comment to obtain sentiment scores, and then conducting various summary statistical analyses.

How we built it

There are three projects in total: Python Project: Responsible for integrating frameworks like TensorFlow, invoking models to obtain sentiment scores.

Java Project: Responsible for calling the Python REST API to retrieve sentiment scores, connecting to TiDB, and providing REST APIs for operations such as writing, querying, and statistical analysis of the dataset.

NodeJS Project: Responsible for frontend display, user interaction, and calling the REST API in the Java project.

Challenges we ran into

If there is a certain level of error between the model's output scores and the actual sentiment, it indicates a need for a higher accuracy model.

Accomplishments that we're proud of

Expanding the exploration of ML applications and successfully implementing the usage of TiDB Serverless.

What we learned

TiDB Serverless indeed offers great convenience by significantly reducing the entry barriers for developers to use TiDB.

What's next for Analyzing stock market movements based on sentiment analysis

To improve the accuracy of sentiment analysis scoring, it is essential to expand the collection of the dataset. By conducting research on the correlation between sentiment scores and stock price movements, valuable insights can be gained.

Share this project:

Updates