Upkar Lidder
Upkar Lidder is a Full Stack Developer and Data Wrangler with a decade of development experience in a variety of roles. He can be seen speaking at various conferences and participating in local tech groups and meetups.
Supercharge your bots with AI
Upkar Lidder, IBM
@lidderupk
ulidder@us.ibm.com
@lidderupk
IBM Developer
http://bit.ly/bot-tips
Set Expectations - platform specific
Bot Basics
@lidderupk
IBM Developer
Leverage existing data & AI
@lidderupk
IBM Developer
INTENT RECOMMENDATIONS
Leverage AI
@lidderupk
IBM Developer
INTENT RECOMMENDATIONS
Leverage AI
@lidderupk
IBM Developer
INTENT RECOMMENDATIONS
Leverage AI
@lidderupk
IBM Developer
INTENT RECOMMENDATIONS
Leverage AI
@lidderupk
IBM Developer
Disambiguation
Handle off topic questions
@lidderupk
IBM Developer
DIGRESSIONS
Handle interruptions
@lidderupk
IBM Developer
HANDLERS
@lidderupk
IBM Developer
Watson Cognitive Services - assistant
@lidderupk
IBM Developer
-
Intent recommendations
- leverage human-to-human logs
-
Digressions
- handle related questions
-
Slots / Handlers
- handle interruptions
-
Complicated flow in the Dialog editor
- jumps and conditions
-
Disambiguation
- multiple intents matched for an utterance
-
Search Skill
-
Discovery Integration
-
-
Service Desk Integration
- Intercom, Zendesk
-
Free preview link
-
good for hackathon demos
-
-
Quick integration with Facebook and Slack
-
Voice Agent
-
STT
-
Watson Assistant
-
TTS
-
-
Call Cloud Functions from inside assistant
-
integrate with other services
-
Bot Effectiveness & KPIs
@lidderupk
IBM Developer
- measure, monitor, and understand the behavior of your bot.
-
understand where your assistant is doing well vs where it isn’t and to potentially focus your improvement effort to one of the problem areas identified.
-
Coverage is the portion of total user messages your assistant is attempting to respond to.
-
Effectiveness refers to how well your assistant is handling the conversations it is attempting to respond to.
-
understand relative performance of each intent and entity, the confusion between your intents, as well as the root cause of utterance issues. This information helps you prioritize your improvement effort.
- generate the confusion matrix, calculate TP, FP, FN and TN, the total number of utterances, average helpfulness, precision and variance, summarize the possible root causes and display the results.
https://ibm.biz/assistant-notebooks
Measure Notebook - Live Bots
Effectiveness Notebook
Thank you
Let's chat !
@lidderupk
IBM Developer
Upkar Lidder, IBM
@lidderupk
https://github.com/lidderupk/
ulidder@us.ibm.com
chatbot-ai-ml-summit
By Upkar Lidder
