Day: March 10, 2023

Natural language processing

Thank you for tuning in to this audio only podcast presentation. This is week 111 of The Lindahl Letter publication. A new edition arrives every Friday. This week the topic under consideration for The Lindahl Letter is, “Natural language processing”

We have reached the 7th installment of the new series for 2023 related to AI. In just a couple of weeks we are going to venture out of these framing topics into the heart of my journey back into polling and sentiment analysis. Toward the end of that journey it will be very clear on the relationships between AGI, AI, and sentiment analysis. Processing to the point of understanding then adding a degree of directionality in terms of sentiment may be beyond the current state of things at this moment. A lot of debate will go into what is understanding, the occurrence of it, and does it have to be sustained to be relevant. Commonly people try to sum that debate up within the context of next best alternative consideration and that is interesting. However, that debate could be summed up within the analogy of trying to argue that a really well drawn map could provide understanding. It certainly could provide the pathing and the next best routing assuming you have a general idea of where you are going and that path is covered on the map. 

Let’s zoom back out for a minute and look at a relatively current webinar. A recent one hour webinar from Stanford online professor Christopher Potts related to GTP-3 & Beyond was pretty interesting. It’s easy to find on YouTube and you can watch it faster than 1x if you enjoy that sort of thing. 

https://www.youtube.com/watch?v=-lnHHWRCDGk

Outside of that consideration, I’m wondering if we are going to end up moving from people writing books and articles to a world where people are certifying books and articles for correctness. Some sort of model is used to produce content and then a person edits it to or just verifies its correctness. People have spent a lot of time trying to figure out how to handle natural language processing (NLP). As editors of that content people could act as the ultimate gatekeeper of knowledge and correctness within what is spit out of the systems. We are after all capable of processing natural language ourselves. It’s just a lot harder to explain or rationalize the rule systems that each individual person uses to achieve that objective. 

Here are 5 scholarly works related to NLP:

Manning, C., & Schutze, H. (1999). Foundations of statistical natural language processing. MIT press. http://www.cs.cornell.edu/home/llee/papers/topost.pdf

Bird, S., Klein, E., & Loper, E. (2009). Natural language processing with Python: analyzing text with the natural language toolkit. ” O’Reilly Media, Inc.”.

Chowdhary, K., & Chowdhary, K. R. (2020). Natural language processing. Fundamentals of artificial intelligence, 603-649. https://strathprints.strath.ac.uk/2611/1/strathprints002611.pdf

Hirschberg, J., & Manning, C. D. (2015). Advances in natural language processing. Science, 349(6245), 261-266. https://nlp.stanford.edu/~manning/xyzzy/Hirschberg-Manning-Science-2015.pdf 

Manning, C. D., Surdeanu, M., Bauer, J., Finkel, J. R., Bethard, S., & McClosky, D. (2014, June). The Stanford CoreNLP natural language processing toolkit. In Proceedings of 52nd annual meeting of the association for computational linguistics: system demonstrations (pp. 55-60). https://aclanthology.org/P14-5010.pdf 

You could go out and watch Christopher Manning’s “Natural language processing with deep learning CS224N/Ling284” 23 video course on YouTube [1]. This one contains a lot of content. 

Maybe you were looking for a more media rich introduction to NLP. The team over at TensorFlow also has a collection of 6 videos on the topic that are going to be a lot more visually involved [2]. They are also considerably shorter than the 23 video course from Chirstopher Manning. 

Rounding this one out will be a video from the IBM Cloud team which was pretty decent. 

Links and thoughts:

https://podcasts.apple.com/us/podcast/kevin-killed-sydney-reddits-c-e-o-defends-section-230/id1528594034?i=1000601415192

Top 4 Tweets of the week:

Footnotes:

[1] https://www.youtube.com/watch?v=rmVRLeJRkl4&list=PLoROMvodv4rOSH4v6133s9LFPRHjEmbmJ 

[2] https://www.youtube.com/watch?v=fNxaJsNG3-s&list=PLQY2H8rRoyvzDbLUZkbudP-MFQZwNmU4S 

What’s next for The Lindahl Letter? 

  • Week 112: Autonomous vehicles
  • Week 113: Structuring an introduction to AI ethics
  • Week 114: How does confidential computing work?
  • Week 115: A literature review of modern polling methodology
  • Week 116: A literature study of mail polling methodology

If you enjoyed this content, then please take a moment and share it with a friend. If you are new to The Lindahl Letter, then please consider subscribing. New editions arrive every Friday. Thank you and enjoy the year ahead.

A golden window of writing adventures

Things got off to a better start today. My energy level was higher at the start of the day. Rocky dog did kick off that day a few minutes before my alarm clock should have gone off. Apparently, the dog based alarm clock was running a little bit early. It’s entirely possible that next week will usher in a golden window of writing adventures. We will see how that one goes. I should have some really good blocks of time to engage in some active writing. That does not always equate to a higher degree of productivity. It just opens the window to it. That opportunity has to be realized to materialize anything. One of the things I want to spend some time working on next week is deeply considering my 5 year writing plan to evaluate where things are and what things should be changed up. 

Throughout this month I have been pretty consistent about sitting down and producing some blog related content. Maybe a bit more of an ongoing narrative would be helpful. Perhaps at the crux of this problem is that each one of these missives is the opening part of getting started. Functionally it is a similar thing that happens every day as the process of going from being awake to being ready to be productive starts. That is what I’m attempting to rationalize anyway. An ongoing narrative would be related to the things that I’m interacting with and an explanation of things that are happening along the way. Probably the only way to really do that would be to make sure that a bunch of starter ideas are seeded in Google Keep throughout the day. At the start of the day instead of writing without a prompt. I really just open a blank page and begin to make my way into the writing process. Things would instead start out by grabbing the previous day’s seeds and beginning to work bringing form to those ideas. 

That method of seeding would certainly produce an ongoing narrative that was essentially about my interactions with things. It would push this from being a more functional journal and into a more reactionary journal. For better or worse that might be a more interesting thing to consider for a couple of weeks. I pretty much did that during the million word challenge where I had a backlog and used that to help focus my writing efforts every morning. During the course of trying to produce 3,000 or more words per day for a blog that is going to be required.