Database – Caption App

In order to have a classification of facial expressions linked to different captions and categories, I have decided to create a database where for each and every facial expression(happy, sad, neutral, anger, fear, surprised) there is a collection of captions with different categories such as: funny, simple, catchy, clever, motivational, sassy, art, quote, lyrics and descriptive.

Initial idea: The initial idea was to data scrape a lot of websites and platforms into a database, however when going through the majority of captions in the websites, they are pretty outdated and mainstream

Current development: I have created a CSV table containing some caption I have scraped from websites and some I have added and deleted manually. I used this formula to extract the lists of captions from the websites directly into the spreadsheet:

For categorising the captions the initial step was to run sentiment analysis on unstructured text (CSV Files) with Monkey Learn. In order for the platform to recognise the category of the caption I had to train the model to meet my specific criteria, by defining what is consider funny, catchy, or clever. Researching and reading more through the system I found out it would not be ideal for my use at is not practical for so many categories which could also be similar for example could be hard to recognise a clever caption with a motivational one, or catchy with funny.

Includes the facial expression, caption corresponding to it, category and resource

This is why I chose to work with a smaller database where the captions were manually categorised by me. There is a total of 455 captions, 9 categories and 9 sources from which the captions where extracted from including:

In order to access the CSV database locally, I had to upload it to the CCI server. Then download it using URLSession.