Click on the image above to go to chatbot page
I found that many dog owners in Victoria are confused about where and when they can let their dogs off leash on beaches. Information signs are hard to find, and fines are high if you’re caught violating the rules.
A primary user would be a dog owner who enjoys traveling and appreciates having the app installed on their phone for easy access. The chatbot could potentially assist with booking pet-friendly flights and hotels, or locating an emergency vet nearby.
The technical components and technology
To build my chatbot, I used IBM Watson, which leverages IBM’s DeepQA software and the Apache UIMA framework implementation. My chatbot is a single-purpose bot focused on performing or automating one specific task or function. It’s designed to provide a fixed set of options for users to choose from, based on what they want to do or the problem they need to solve (Microsoft, 2022).
I also researched two other platforms that seemed suitable:
Landbot — I liked its intuitive drag-and-drop conversation builder. A major advantage is its ability to deploy conversations across the web, WhatsApp, Facebook Messenger, or any API-enabled channel.
Chatbot.com — I appreciated its chat widget, which can be launched on any website within a few clicks. It’s a powerful tool offering extensive customisation options in design, analytics, and user management.
The Challenges and Solutions
The mind map below illustrates how the bot’s conversation can progress. It learns quickly by gathering analytics data. From the backend system, it’s easy to identify mistakes or areas for improvement. One issue I discovered while reviewing users’ questions was that the keyword “dog_beach” had mistakenly been assigned to the wrong intent. This confused the bot and prevented it from providing correct answers. Once the keyword was removed, the problem was resolved.
To give the chatbot more personality, I added images of dogs from various locations around Melbourne. One user testing the app appreciated this, commenting: “Pictures add colour and interest and are on topic.”
A key challenge was delivering accurate answers based on the user’s location. From a development standpoint, embedding the bot into an app presents its own complexities and will require collaboration between developers, UX designers, and product teams.
For projects like this, a strict design system and style guide must be integrated. The biggest challenge is maintaining consistency with brand guidelines and component styles throughout the user experience.
Here is an example of how components should be created using the latest iOS design standards:
USER TESTING
32 people tested the chatbot
It was created with 13 intents only
In the backend analytics I could find other more controversial questions
The bot was able to cover 79.9% of questions and it keeps learning!
The chatbot was able to answer even misspelled words, that weren't included in the backend system.
Thanks to IBM Watson analytics I was able to spot problems like here: #dog_beach was included in #dog_hotel and confused the chatbot
REFERENCES
Wikipedia. (2022). Watson (computer). Accessed 17 March 2022, from https://en.wikipedia.org/wiki/Watson_(computer)
Microsoft. (2022). What is a chatbot? Accessed 17 March 2022, from https://powervirtualagents.microsoft.com/en-us/what-is-a-chatbot/
Landbot. (2022). The Most Powerful No-Code Chatbot Builder. Accessed 17 March 2022, from https://landbot.io/
Chatbot.com. (2022). An all-in-one platform to build and launch conversational chatbots without coding. Accessed 17 March 2022, from https://www.chatbot.com/
Today Digital. (2018). Stronger Insights, Greater Efficiencies – IBM Watson. Accessed 17 March 2022, from https://www.cxtoday.com/analytics/ibm-watson-review/
Chris & Chris on Chatbots. (2016). Build a chatbot in 6 minutes with IBM Watson Conversation [Video]. Accessed 17 March 2022, from https://www.youtube.com/watch?v=MTCc4d-RXP0&ab_channel=Chris%26ChrisonChatbots
Adobe Stock Images. (2022). Chatbot application images. Accessed from https://stock.adobe.com/uk/search?filters%5Bcontent_type%3Aphoto%5D=1&filters%5Bcontent_type%3Aillustration%5D=1&filters%5Bcontent_type%3Azip_vector%5D=1&filters%5Bcontent_type%3Avideo%5D=1&filters%5Bcontent_type%3Atemplate%5D=1&filters%5Bcontent_type%3A3d%5D=1&filters%5Bcontent_type%3Aaudio%5D=0&filters%5Binclude_stock_enterprise%5D=0&filters%5Bis_editorial%5D=0&filters%5Bfree_collection%5D=0&filters%5Bcontent_type%3Aimage%5D=1&k=chatbot+application&order=relevance&safe_search=1&limit=100&search_page=4&search_type=pagination&get_facets=0&asset_id=257108384
Thanks, Woof Woof!
