Created by Ashleigh Faith: Focusing on knowledge graph, taxonomy, machine learning, and other data science topics, and making them fun and relatable.
For instance, ever wonder how Amazon can predict what you are interested in? How can LinkedIn connect people from such dispersed disciplines? How does Uber optimize their pickup routes? How does Google or Siri know what I mean? More importantly, how can I organize my Instagram photos?!? It's not magic, its science and while I will not proclaim to know all there is to these questions, but on this channel we will walk through the method to the madness :-)
Thank you, Ashleigh. concept of a Wisdom Graph its an intriguing evolution of knowledge representation that could significantly enhance how we capture, understand, and utilize personal and collective wisdom in the future
Thank you! It’s such a personal thing, knowledge. We all have a different fabric of experience and backgrounds that even the “facts” in a traditional knowledge graph only make for the foundation. It should be interesting to see how things go from here!
Thank you! Yah, I find Paul’s perspective always grounded and insightful. He cuts through the hype and is realistic about what things can and can’t do. Happy you enjoyed it!
Graph databases will be vital in a global platform for collective intelligence. The platform will keep humans in the loop through a conversational interface that merges public parts of conversations into a shared graph representation. It will report on which issues are discussed by the most people, aggregate measures for sentiment about proposed solutions, and show how those measures of sentiment change over time. One big challenge is how the platform will authenticate the identities of humans at the same time that it allows people to share selected parts of conversations anonymously.
I am glad to be more aware of ArangoDB. LLMs within agentic workflows, and adapted to act as interfaces to databases may converge to collective intelligence.
Ha, yah that’s the issue when loading videos sometimes. My copy has the summary timestamp correct but RU-vid probably condenses the video a bit. The summary is at the very end though, I checked! I hope you enjoy it!
Thank you, Ashleigh. This video was very well presented. I do not want to sound too geeky, but personally, I found it moving. While I had heard of the book Personal Knowledge Graphs: Connected Thinking to Boost Productivity, Creativity, and Discovery, after watching your interview with the authors, I knew I needed to purchase it. I was looking for a version on Audible but saw there was only a Kindle version, so I put it on my reading list to read when I have time. Ironically, this video was next on my watch list. To say the least, it was a pleasant surprise. Thank you again 🤗
Thank you! I was trying to harness the ebb and flow of knowledge and its history a bit when I wrote this. I have an article I am writing in a similar vein looking at all the “futuristic” libraries written in the height of sci fi in my opinion in the 30-50s. It’s so facilitating to see how those predictions played out and what it says about us now and then. Thanks for watching and I hope you find the book interesting, if not a bit eclectic!
Whenever I hear "it depends" when I ask a question, it's a sign to me that they're an expert! If someone loses interest in the conversation after you say that, then they didn't really want to learn anything and they had some kind of ulterior motive for asking in the first place lol. Great interview and content, as always!
I love this!!! Thank you. You saved me hours. Quick question, you mention that the triples are generally built on URIs. Does that mean, each URI represents a specific concept or entity?
So glad you found this helpful! Yes, each URI represents a specific concept. It’s the ID, or another way of thinking about it, it’s like the key-value pair of a piece of information in a database.
Well to be fair, this was one of my first videos and it was four years ago. Maybe try a different video and see how you like it. Or maybe my style isn’t your cup of tea. In which case, wishing you well and good luck on the rest of your journey!
@@AshleighFaith It's my pet peeve with NPR as well. I like the content but they are filling the airwaves with young reporters or journalists with cases of this increasingly ubiquitous speech 'affect'. You know more about taxonomy than I do, but gosh.
@GilbertNeal I try to keep my videos conversational and that can creep in at times. The 10 min or less series I have to script to keep the time limit and those don’t have the “speech issue” as much. My new LinkedIn learning course is also scripted so it is tighter.
Hello @AshleighFaith Following on from the last comment, and understanding this video is 3 years old, I wonder if you have any thoughts about where TerminusDB stands now, and its capacity to grapple with the inference complexities of OWL. Really appreciate your thoughts on this!!!!!! Thanks for the excellent videos, as always!
Thank you for the kind words! And yah, I’m sure there are other standards out there that I don’t know about. There are a ton that are specific to the industry the data is from. Would love to see what others are out there I may have missed in this video!
@@AshleighFaith Yeah that is the problem, there seem to be too many metadata standards to choose from :) I am curious as to how to evaluate my choice of a metadata standard for a particular data. Maybe could be a topic for another video from you :)
@shelf557 I love that idea! It’s tricky when there are some staples to choose from like do you use SKOS definition, RDF definition, or dc:description? They all can be used for the same metadata. I try to keep within the same standard if I can for consistency and I don’t use standards that are not supported or used much anymore because that means they are brittle, but outside of that it comes down to your use case but that’s what I can cover in the video. Thanks again for the great suggestion! If you want me to give you a shout out in the video, and you don’t mind sharing your first name at least, I will add that to the video as a way of thanks.
Thanks for producing this video. I did not realize that GraphDB was so advanced. This feature is very relevant since I have a process to turn all of my data into RDF/TTL. I now wonder how GraphDB compares to Neo4j and AWS Neptune. I have a dataset that updates monthly that is roughly 700MB in Neo4j. It would be interesting to compare technologies.
I know Neo just came out with a csv to graph library and they also have neosemantics that converts to RDF. Neptune doesn’t have a native data science studio but they obviously have sagemaker and all the other ML tools that you can use in their suite. And yah, I have wanted to do a side by side with the same dataset and queries over all the graph databases I have reviewed but I just haven’t found the time yet!
@@AshleighFaith Hi. In my situation, the source files are compressed S3 objects. Our infrastruture policy did not allow creating a large temporary EBS volume to hold a copy of the data. Therefore, I wrote a python script to read the files in S3, then pump the data into Neo4j. This approach traded local volume size for import speed. This comment is apropos of nothing. I just wanted to vent. :)
While I would definitely love a copy of Rick’s book, I am in Australia so postage would be prohibitive. I also have found a great deal on Elsevier for both the physical book and a digital copy.
I have just found your channel and I am finding your videos so helpful. So many of them are exactly what I have been looking for. Your presentation and style is engaging and makes it easier to get my head around the terms and their uses. Well done. Off now to watch a few more of your thesauri videos.
Those are definitely characteristics of a taxonomy, but the key is what you will use it for. Taxonomies are used to classify and then organize assets into a browse tree, or to help search find content for retrieval
Hi there! Great video on content and data modeling. I noticed your channel has 2.98k subs, but this video has only 88 views in 5 days. Incorporating some SEO optimization techniques in your title, description, tags, and updating your channel settings could significantly improve your reach and engagement. I'd love to help you with this and boost your channel's performance. Feel free to reach out! Best, Rezaul Bari RU-vid SEO Expert
Excellent conversation. Thank you! Summarising the key questions one must think of when modeling. Then thought i'd flip these into what this means to authors. What is the content for? / How will it be used? What do you need to model? What rules and validations do you need? How do you balance granularity with flexibility? Does the authoring interface support content quality? Flip this into the author’s perspective: a content model (and CMS) should meet these criteria: * Content is for < these audiences > in < this format > published to < this channel > * I need to < author > these fields and < pull/reference > content for these fields from our DAM * Authoring is of higher quality/with fewer errors because we are < re-using > content/data, * There is content for all < mandatory fields > so I’m able to successfully author * CMS < roles and permissions > reflect the work I need to do
All great questions! And ones that can sometimes take a while to answer. I have found a lot of folks have a lot of data, but they don’t know what it’s there for or from and therefore are scared to do anything with it. It’s like a scavenger hunt, wrapped in a detective story, wrapped in a ghost story with all the skeletons in closets, sleuthing to get an understanding of the data and content, and then having to go search for the systems and people that use or own it, if they even exist.
Oh that’s such a good point! The depth seems to correspond to maybe the subtlety of the data and those subtleties are more volatile? It’s a good point to consider though. Thanks for adding this to the conversation!
\m/ love this (and your content in general), maybe almost as much as I love Metallica \m/ Thank you thank you thank you from a linguist phd student trying to use ontologies to create/automate synthetic datasets <3 I feel like a real-life semanticist often after watching your stuff :)
Eeeeexactly what i am doing. Not only data but also information. Curation sophistication is what matters. Anyway for now i want my own great data for my LLM in the future. The way to go))
Oh that makes me think of another good video idea. What’s the difference between data and information. I use them interchangeably, mostly for convenience and simplicity, but I also know lots of folks delineate between the two. Seems like a good video idea to me!
@@AshleighFaith ontology and knowledge graph are few ways to process data and turn it in to information. If you can expand different ways of data processing and also INTERPRETING (we need insights not just information right?))), this can make a great vid. And also if you can, add more charts and graphics or models and schemes. And thank you very much for your work
@@emotionalmindedstate this is great feedback, thank you! When you say add more charts, graphics, models and schemes, are you meaning just more visual aids, or would you like for me to cover more types of schema and models? I do try to add in visuals whenever I can, but they do take a very long time to create so I try to make them count when I use them. For this video, its a data as a product so its difficult to show what that looks like without just showing a dataset on screen. For other videos, I get your point!
@@AshleighFaith you can draw em by hand in a paint or else. We will forgive you))) And yes. I can speak for myself. I d like to see BOTH. And visuals and schema models covers )))
That is a great question! While SKOS is a very common linked data standard, Google and other search engines have preferred standards they use, which are listed in the video. SKOS would be used as an internal schema for your vocabularies to be stored and maintained as a controlled list. The schema in this video would then use those SKOS tags in the search engine schema fields for describing what the webpage is about. That said, Google also recognizes JSON-LD, where you could encode your SKOS tags as JSON-LD fields instead of schema.org. Hope this helps!
Technically great, but I think that current curation models (across multiple domains) are federated. Operationally this requires a large curator user base with intermittent (and in frequent use). The pricing model of Graphologi makes this un affordable in a federated curation operating model
My entire channel is dedicated to building out graphs and how to use them so if you would like a demo on how to populate the ontology, check out some of my other videos. Thanks for stopping by!
Hello! I've been commenting on your content but im not seeing my comments. I would love to connect with you as it relates to my project! I would love to connect you with the OriginTrail eventually.
Hey! So strange. This is the first comment I have seen and I don’t see any held for review. Well this one got through! My email is in the description box if you want to send me details.
@@AshleighFaithawesome! I’ll be attending the knowledge graph conference today with the co-founder of OriginTrail and Bob Metcalfe! I’ll certainly be reaching out. So excited to connect you with the team.
I found this video extremely relevant, insightful, and practically useful - today. And, I have to say that the interplay between you and Jörg was so refreshing as you geeked-out about the connectedness of data! Thank you!
