OHA Chatbot - Team Name 404

Inspiration

OHA works to improve the well-being of Native Hawaiians through advocacy, research, community engagement, land management, and the funding of community programs. OHA also offers a wide range of services, such as grants, scholarships, loans, Hawaiian Registry, legal consultation, etc. The problem is that OHA receives 13,000 calls annually. 11% of the calls never reach anyone and 30 percent of the calls require transfers. Therefore, with the help of a chatbot, users can have a higher chance of finding what they are looking for without waiting on a long waitlist or ending up calling the wrong department.

What it does

Our chatbot aims to provide users with accurate information and links by extracting keywords from their messages such as "jobs at OHA", which will provide the user with information on how to apply for jobs at OHA. Or the user could type in "grants" and the chatbot will provide appropriate links and information regarding grants the OHA is supporting.

How we built it

Powered by Google's Dialogflow API, which is a natural language understanding platform. This is where most of the AI/ML training and magic happens that makes our chatbot chatty. We then integrate our chatbot using third-party software called Kommunicate that effortlessly integrates Dialogflow into our website with a simple script. Kommunicate is a help desk service, which will allow OHA employees to reach out to people using the chatbot.

Challenges we ran into

Our biggest challenge was our chatbot intents (the keyword that our chatbot looks at in order to give the correct response). Some intents were too similar to each other such that the chatbot gives an incorrect response. On top of that, even though we have tried to cover as many areas of the OHA website, there are still many more intents to implement to make our chatbot cover everything you can find online.

Accomplishments that we're proud of

We are just very grateful and proud of the opportunity to learn something new and possibly make an impact on native Hawaiians by making a difference in their lives.

What we learned

We learned to use Dialogflow to create an agent and provide intents to train the AI to recognize the intent (what the user is asking) even when it is phrased differently. We also learned how to integrate a voice bot so users can make a phone call if the web is unavailable or is more inconvenient compared to the web version.