Ideation Feature Week

[tanishapatil1234]

BACKEND TICKET

The following is a description of the work I (Tanisha Patil) have completed for our group passion project.

1. Backend Development

• Created the Quote Java Class

2. CRUD Functions

• Developed all CRUD functions

  

• Testing them in PostMan

  Create

  

Read

Update

Delete

3. AI Development

• I have been experimenting with different model types (Logistic regression, CNN, etc. ) I have decided on a CNN type but came to the realization quickly that it would be too time consuming to start from scratch. This is why I am using a preexisting CNN template model and fitting it to preprocessed facial image data. Here is a code snippet of what I have been coding so far. This is far from done and will be my main focus for this week and the next. Plan to have tangible model in two weeks time.

[tanishapatil1234]

FRONTEND TICKET

• Live demonstration of deployed frontend

[isabelle926]

Feature: Your AI Therapist

Therapy is expensive, but using an AI chatbot is (mostly) free! For this week, I added a chatbot that acts as a therapist to users.

Frontend:

My commit can be found here.

On the home page (and other pages), there is a button that users can press to open a chat field.

My code!

This opens up a window that is connected to the backend.

My code!

Backend:

My commit can be found here.

For the backend, I decided to use Apache OpenNLP library, which is a machine learning toolkit that can process natural language text. Using an NLP is useful for our purposes since it allows for "communication" between a human and a machine. I decided to use Apache OpenNLP in particular because it's particularly useful for categorizing text into "human" categories, such as emotions.

This process is known as lemmatization! Lemmatization is the process of grouping together different inflected forms of the same word. My chatbot goes through the following steps:

• Train categorizer model with latest samples data.
• Take input from user (through console).
• Break sentences.
• Tokenize each sentence into words.
• Find POS tags of each words as its required in next step.
• Lemmatize each word using tokens & POS tags. This will make it very easy to categories as we don’t have to have all lemma possibilities in categorizer samples data.
• Categorize lemma tokens & then find answer for detected category.

What next?

• [ ] Connect frontend to backend
• [ ] Improve CSS of frontend
• [ ] Get emotion detected from camera to automatically open chatbot and print responsive message
• [ ] Consider using AIML instead of OpenNLP
• [ ] Code backend in python instead