This process is designed to work with a variety of product review systems, as the requirements for the data input are simply that the data file be in either csv or json format and that it have a column or field that can be identified for either product id or product name, a column or field that represents timestamp and one that contains product ratings. In order for the process to use sentiment analysis, the data must also contain the product review text.
The columns and fields do NOT need any specific naming structure, as you are able to define that at the start of the process.
You are given two outputs, one is a sentiment analysis score for each data point that contains product review text, and finally an approximately 200 to 500 word product insights write up for each of the top three most reviewed items in the dataset (please see the youtube video for instructions on how to add additional product insights, if you would like it to return more than three).
Please note that under the current system, the product insights are not preserved anywhere after the final form is closed. You will need to be prepared to copy and paste any insights you would like to save into another program.