Closed toddysm closed 2 years ago
Steven and I worked on this issue together and adjusted the scope a little bit. We are using a sample of less than 100 sentences, but the work we have done could be extended.
We also decided to utilize work done by other classmates with the Vader natural language processing to classify feedback as positive, negative, or neutral.
Initial time estimate (each, me & Steven) ~10hrs
Next issue would be to iterate on the existing script developed by Eric to generate the output into a JSON format, so it may be consumed by the front end client.
Further research on ML classification, different types of sentiment analysis are also issues we can explore.
Create a simple file with 100 sentences Give a random number (weight) between 0 and 10 to each sentence. Use this as input for ML algorithm that classifies each sentence that has wight more than 2 as good and all others as bad.