Intelligent Chatbots to Combat Depression in Students

Intelligent Chatbots to Combat Depression in Students

As students, we have all gone through situations where we totally had no idea on how to deal with emotional attacks. May it be the pressure that we experience due to the tight schedule or any other reason for that matter, we all have found it really difficult to deal with such issues, and to add more at times it was also difficult to identify as of whom to approach for the right advice. Taking into consideration the tender age of the students it also becomes very important to guide them on the right path.

And this time, the field of artificial intelligence would probably be able to help out with this issue.
A team of researchers has proposed the idea of introducing a chatbot that could help the students deal with issues faced and could suggest solutions to their problems.
Now, What is a Chatbot??

A computer program that conducts a conversation with humans either in voice or textual method is commonly called a chatbot. Chatbots are basically classified into two types. The first one is based on a fixed set of hand-crafted rules where the chatbot would reply according to previously mentioned regulations and the second one is an intelligent bot. Apple Siri, Google Allo are some currently available chatbots and they use Natural Language Processing (NLP) which provides the machine with the ability to allow communication between machine-to-user and user-to-machine using human natural language.


The Proposed System

Here, the researchers propose an intelligent social therapeutic chatbot where the bot would ask a few questions to the user and would understand the problem. It then distributes the text into emotion labels ie; Happy, Joy, Anger Shame, Fear, and furthermore and would analyse the mental state of the person using the chat data. Further, it calculates the percentage of negativity in the chat and with the help of the negative content in the chat, it would extract the emotion of the person.

For detecting the emotion of the person, the researchers deployed three popular deep learning classifiers namely, Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), and Hierarchical Attention Network (HAN).

Convolutional Neural Network (Conv Net or CNN): A category of artificial neural network which has proven its efficiency in the field of image recognition and data classification.

Recurrent Neural Network(RNN): Recurrent Neural Networks are used to process variable-length sequences of inputs. Hence, this becomes applicable for tasks such as connected handwriting recognition and speech recognition.

Hierarchical Attention Network (HAN): This Artificial Neural Network is used to highlight the informative words in a sentence.


Results

Using the emotion label, the neural networks will classify the mental state of the user and then advises to take the mental treatment as follows:

  • Zero depression- No therapy requirement. The user is completely fine and requires no treatment.
  • Slightly stressed- Relaxation required to shed stress. The user is mildly or occasionally stressed. She/he may need irregular breaks to cope up with stress.
  • Highly stressed- Reduce stress in life. The user is mildly stressed and requires regular breaks between works to shed the accumulated stress.
  • Slightly depressed- Engage in recreational activities. The user is moderately stressed and on the borderline, i.e., he/she may be on getting highly stressed. Meditation, relaxation is the need of the hour.
  • Highly depressed- Engage in recreational activities. The user is highly stressed. Meditation, relaxation is the need of the hours. Possibly needs to meet a doctor.
  • Here, in the proposed methodology, the chatbot through interaction would try to prevent the pessimistic actions and rebuild more constructive thoughts.

Reference: IEEE

Have a look at our recent article: Detecting long-hidden microquakes using AI

Author