Reference no: EM132290553 , Length: 20 pages
The purpose of this assignment is to illustrate the application of cognitive computing and machine learning concepts through building and training a chatbot that simulates the human conversation for the defined scenarios. You will use IBM Watson Assistant service in IBM Cloud for chatbot implementation.
The assignment due date cannot be changed because the work cannot be accepted after the last day of class.
Before working on this assignment
• Complete all required reading in the week 10-12 course content
• Complete the chatbot self-assessment in the week 10-12 course content
• Complete the chatbot walkthrough in assignment 4 folder, including ungraded exercises
• Make sure that your IBM Cloud account is active
Assignment
• A single Microsoft Word or PDF file with a report that addresses all requirements above and all bullet points in the grading rubric.
• The JSON file. See the chatbot walkthrough for instructions on how to export the JSON file.
Only one assignment submission is allowed. Make sure to attach 2 files before hitting the submit button.
To get credit for assignment deliverable, the discussed application and scenarios must be original. The chatbot implementation must meet the defined purpose.
Follow the outline below to write a detailed report on your experiment.
1. Introduction
• Define the purpose of the chatbot and the conversation requirements. How will an end-user interact with the bot? What are the expected benefits of the chatbot you are going to build?
• Use Visio or draw.io build the dialog flow diagram. Discuss at least 5dialog scenarios.
2. Implementation
• Use Watson Assistant service to build the chat bot per requirements above. The dialog implementation is required to utilize
o User entities and system entities
o Each value in the user entity should have at least 3 synonyms
o Intents; each intent should have at least 5 examples
o context variablesand slots to collect and store user input
o handlers to answer quick questions while collecting information from a user.
o regular expressions to parse user input
o conditional responses
o the responses at each parent node should have at least 3 variations
o use image, option, and pause response types in addition to text response type.
• Discuss the development steps, including the entities and intents you created, the application of context variables and slots, the dialog nodes and flow.
• Use the try it out panel to testthe dialog scenarios disused above. Include the screenshots that show the bot responses and your input. Explain and interpret how the bot handles and responds to your input.
• Use the try it out panel to Illustrate handling the unexpected input and training the bot. Include the screenshots of the dialog. Explain how the bot handles your input and the approach you use to train the bot.
3. Virtual Assistant and Conversation Data Analysis2-days experiment
• Create a virtual assistant. Share the public link with 2-5 friends and/or family members.
• Day 1
o Ask each participant to run 3-4 conversations.
o At the end of the day, take screenshots of the conversation analytics overview tab. Include the screenshots in your paper and interpret the metrics and plots.
o Check the interactions between user and chatbot on the user conversation tab and identify the messages with unrecognized intents and with misclassified entities. Adjust the classification as needed and discuss the adjustments in your paper.
• Day 2
o Ask each participant to run the same 3-4 conversations again.
o At the end of the day, take screenshots of the conversation analytics overview tab. Make sure that the time period includes day 1 and day 2.
o Include and discuss the screenshots of the overview page in your paper. Discuss the changes in top entities and top intents during day 2.
o Check the interactions between chatbot and user and discuss the findings in the paper.
o If you were going to run this experiment for a third day, what additional entity and intent classification adjustments would you make?
4. Chatbot Enhancement
• What approaches did you use to personalize the interaction with the chatbot?
• Choose 2 Watson services from the list below and discuss how would you integrate them to enhance the chatbot's understanding of user's input and to improve the relevancy of responses.
• What messaging platform would you use to deploy the chatbot? Explain the rationale for your platform choice.
• What approaches would you use to further train the chatbot after deployment
5. Conclusion.
• Discuss 3 chatbot implementation challenges and at least 3 recommendations to build the "smart" chatbots.
• Would the chatbot that you implemented replace the interaction with the live person? Why or why not?
Attachment:- Assignment Requirements.rar