DEVELOPING CONVERSATIONAL INTERFACES FOR IOS
2 likes240 views
The document discusses the development of conversational interfaces for iOS, focusing on natural language understanding and user intent detection. It covers tools such as SiriKit, Dialogflow, and Core ML for implementing speech recognition and NLP in mobile applications. The author, Martin Mitrevski, shares insights on training agents, challenges like multi-language support, and the importance of user adoption.
DEVELOPING CONVERSATIONAL INTERFACES FOR IOS
- 1. DEVELOPING CONVERSATIONAL INTERFACES FOR IOS Martin Mitrevski 1
- 2. ABOUT ME ➤ From Macedonia (7192 km from Boston) ➤ iOS at Netcetera ➤ Blogging on martinmitrevski.com ➤ Book Author - “Developing Conversational Interfaces for iOS” ➤ @mitrevski 2
- 4. NATURAL LANGUAGE UNDERSTANDING CONCEPTS I would like to eat a Boston Cream Pie. Intent Entity I’ve heard it’s delicious. Context 4
- 6. ADDING CI TO YOUR APPS training an agent with domain knowledge provide an interface to a mobile app speech recognitio n on the mobile device intent detection and entities extraction executing the user’s request 6
- 7. WHAT CAN YOU USE?
- 8. SIRIKIT 8
- 9. SIRIKIT 9
- 11. SIRI SHORTCUTS ➤ Expose capabilities of your apps to Siri
- 12. DIALOGFLOW 12
- 19. DEMO 19
- 20. WIT.AI 20
- 21. WIT.AI 21
- 22. WIT DELEGATE PREVIEW TEXT 22
- 23. WIT DELEGATE INTENT DETECTION 23
- 24. CORE ML 24
- 25. CORE ML SENTIMENT ANALYSIS 25
- 26. CREATE ML
- 27. TF-IDF ALGORITHM Term in document Term in other documents Term frequency * inverse document frequency Importance of word 27
- 29. WORD COUNTING 29
- 30. DEMO 30
- 31. NATURAL LANGUAGE FRAMEWORK ➤ Language identification ➤ Tokenization ➤ Named entity recognition ➤ Custom NLP models (Text classification and Word tagging) ➤ Works well with Core ML and Create ML
- 32. CHALLENGES ➤ Multi-language support ➤ Offline vs. Online (privacy) ➤ It will never be perfect ➤ User adoption 32
- 33. THANK YOU! ➤ Questions? ➤ martinmitrevski.com ➤ @mitrevski 33