AI for Libraries
0 likes503 views
The document discusses artificial intelligence (AI) in the context of libraries, presented by Bohyun Kim during the 23rd National Convention on Knowledge, Library, and Information Networking. It covers AI's capabilities, how it works, and various applications specific to libraries including information retrieval and chatbots. The presentation also addresses the future of AI, its applications in content processing, and encourages collaboration between libraries and AI tool developers.
![ThankYou!
Bohyun Kim
@bohyunkim [Twitter]
http://bohyunkim.net/blog [Blog]
CTO & Associate Professor,
University of Rhode Island Libraries,
USA
Publications:
http://www.bohyunkim.net/blog/publications
Presentations:
http://www.bohyunkim.net/blog/presentations
Slides:
https://www.slideshare.net/bohyunkim](https://image.slidesharecdn.com/delnetbkim-201005181418/85/AI-for-Libraries-49-320.jpg)
AI for Libraries
- 1. AI for Libraries DELNET (Developing Library Network) Annual Lecture - 23rd National Convention on Knowledge, Library and Information Networking (NACLIN), Online(/India), Sep. 23, 2020. Bohyun Kim CTO & Associate Professor, University of Rhode Island Libraries, USA
- 3. What Comes toYour Mind WhenYou Hear “AI”?
- 7. Malicious AI
- 8. AI for Libraries I. What CanToday’s AI Do? II. How Does AIWork? III. Deep Learning Applications IV. AI for Libraries V. Q/As (15 min)
- 9. What is the Purpose of AI?
- 10. I.What CanToday’s AI Do?
- 12. https://www.nytimes.com/2019/02/05/busines s/media/artificial-intelligence-journalism- robots.html AITools for NLG • Heliograf, Washington Post • Wibbitz, USAToday • Cyborg, Bloomberg News
- 13. ComputerVision real-time translation and object identification from the camera screen
- 14. More Examples of AI Applications • Siri • Alexa • Tesla • Amazon • Netflix • Pandora • Nest
- 16. Madhumita Murgia, “AI’s New Workforce:The Data-Labelling Industry Spreads Globally,” FinancialTimes, July 23, 2019, https://www.ft.com/content/56dde36c-aa40- 11e9-984c-fac8325aaa04.
- 17. II. How Does AIWork?
- 19. Symbolic AI is Rule-Based. (50’s – 80’s) https://medium.com/@sunilpnwr/expert-systems-42715a5a5b14
- 20. Machine Learning is Data-Driven. Diagrams from Francois Chollet and J. J. Allaire, Deep Learning with R, 1st edition (Shelter Island, NY: Manning Publications, 2018).
- 21. https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/ Deep Learning Uses Hidden Layers.
- 23. https://www.nytimes.com/2012/06/26/technology/in-a-big-network-of-computers-evidence-of-machine- learning.html • +1 billion parameters • 16,000 CPU cores • 10 million unlabeledYouTube videos
- 24. III. Deep Learning Applications
- 25. • 9-layer neural network • +120 million weights • 4 million images https://research.fb.com/publications/deepface-closing-the-gap- to-human-level-performance-in-face-verification/ https://www.technologyreview.com/s/525586/facebook-creates- software-that-matches-faces-almost-as-well-as-you-do/
- 26. https://www.nature.com/news /deep-learning-boosts- google-translate-tool-1.20696
- 31. IV. AI for Libraries
- 32. (i) Abstracting & Indexing
- 34. (ii) Information Discovery / Retrieval & Research Insight
- 38. (iii) Feature Detection & Content Extraction
- 39. https://2018.code4lib.org/talks/deep-learning-and-historical-collections Also see : Eric Phetteplace, Bohyun Kim, and Ashley Blewer, “Reflections on Code4Lib 2018,” ACRLTechConnect (blog), March 12, 2018, https://acrl.ala.org/techconnect/post/reflections-on-code4lib-2018/.
- 41. (iv)Voice User Interface (VUI) & Chatbots
- 46. JessamynWest, “TILT #55 - ‘Hey Google AREWomen Smarter than Men...?,’” TinyLetter (blog), accessed June 11, 2018, http://tinyletter.com/jessamyn/letters/tilt-55-hey-google-are-women-smarter-than-men.
- 47. Consider: How can libraries / content providers / information system vendors make it easier for AI- powered tools to accurately ingest, process, and evaluate the content and the metadata?
- 49. ThankYou! Bohyun Kim @bohyunkim [Twitter] http://bohyunkim.net/blog [Blog] CTO & Associate Professor, University of Rhode Island Libraries, USA Publications: http://www.bohyunkim.net/blog/publications Presentations: http://www.bohyunkim.net/blog/presentations Slides: https://www.slideshare.net/bohyunkim