An overview on the history of ideas leading to knowledge graphs, with plenty of references. Useful for anyone who wants to understand the background of the field, and probably the best current such overview.
“Look, I’ll be honest, if living in the US for the last five years has taught me anything is that any government assemblage large enough to try to control a big chunk of the human population would in no way be consistently competent enough to actually cover it up. Like, we would have found out in three months and it wouldn’t even have been because of some investigative reporter, it would have been because one of the lizards forgot to put on their human suit on day and accidentally went out to shop for a pint of milk and like, got caught in a tik-tok video.” -- Os Keyes, WikidataCon, Keynote "Questioning Wikidata"
It is wonderful to live in the Bay Area, where the future is being invented.
Sure, we might not have a reliable power supply, but hey, we have an app that connects people with dogs who don't want to pick up their poop with people who are desperate enough to do this shit.
Another example how the capitalism that we currently live failed massively: last year, PG&E was found responsible for killing people and destroying a whole city. Now they really want to play it safe, and switch off the power for millions of people. And they say this will go on for a decade. So in 2029 when we're supposed to have AIs, self-driving cars, and self-tieing Nikes, there will be cities in California that will get their power shut off for days when there is a hot wind for an afternoon.
Why? Because the money that should have gone into, that was already earmarked for, making the power infrastructure more resilient and safe went into bonus payments for executives (that sounds so cliché!). They tried to externalize the cost of an aging power infrastructure - the cost being literally the life and homes of people. And when told not to, they put millions of people in the dark.
This is so awfully on the nose that there is no need for metaphors.
San Francisco offered to buy the local power grid, to put it into public hands. But PG&E refused that offer of several billion dollars.
So if you live in an area that has a well working power infrastructure, appreciate it.
Sorry for showing off, but it is just too cool not to: here is a visualization of my academic lineage according to Wikidata.
"Bring me to your leader!", the explorer demanded.
"What's a leader?", the natives asked.
"The guy who tells everyone what to do.", he explained with some consternation.
"Oh yeah, we have one like that, but why would you want to talk to him? He's unbearable."
September 24 was the AKTS workshop - Advanced Knowledge Technologies for Science in a FAIR world - co-located with the eScience and Gateways conferences in San Diego. As usual with my trip reports, I won't write about every single talk, but offer only my own personal selection and view. This is not an official report on the workshop.
I had the honor of kicking off the day. I made the proposal of using Wikidata for describing datasets so that dataset catalogs can add these descriptions to their indexes. The standard way to do so is to use Schema.org annotations describing the datasets, but our idea here was to provide a fallback solution in case Schema.org cannot be applied for one reason or the other. Since the following talks would also be talking about Wikidata I used the talk to introduce Wikidata in a bit more depth. In parallel, I kicked the same conversation off on Wikidata as well. The idea was well received, but one good question was raised by Andrew Su: why not add Schema.org annotations to Wikidata instead?
After that, Daniel Garijo of USC's ISI presented WDPlus, Wikidata Plus, which presented a prototype for how to extend Wikidata with more data (particularly tabular data) from external data sources, such as censuses and statistical publications. The idea is to surround Wikidata with a layer of so-called satellites, which materialize statistical and other external data into Wikidata's schema. They implemented a mapping languages, T2WDML, that allows to grab CSV numbers and turn them into triples that are compatible with Wikidata's schema, and thus can be queried together. There seems to be huge potential in this idea, particularly if one can connect the idea of federated SPARQL querying with on-the-fly mappings, extending Wikidata to a virtual knowledge base that would be easily several times its current size.
Andrew Su from Scripps Research talked about using Wikidata as a knowledge graph in a FAIR world. He presented their brilliant Gene Wiki project, about adding knowledge about genes and proteins to Wikidata. He presented the idea of using Wikidata as a generalized back-end for customized frontend-applications - which is perfect. Wikidata's frontend is solid and functional, but in many domains there is a large potential to improve the UX for users in specific domains (and we are seeing some if flowering more around Lexemes, with Lucas Werkmeister's work on lexical forms). Su and his lab developed ChlamBase which allows the Chlamydia research community to look at the data they are interested in, and to easily add missing data. Another huge advantage of using Wikidata? Your data is going to live beyond the life of the grant. A great overview of the relevant data in Wikidata can be seen in this rich and huge and complex diagram.
The talks switched more to FAIR principles, first by Jeffrey Grethe of UCSD and then Mark Musen of Stanford. Mark was pointing out how quickly FAIR turned from a new idea to a meme that was pervasive everywhere, and the funding agencies now starting to require it. But data often has issues. One example: BioSample is the best metadata NIH has to offer. But 73% of the Boolean metadata values are not 'true' or 'false' but have values like "nonsmoker" or "recently quitted". 26% of the integers were not parseable. 68% of the entries from a controlled vocabulary were not. Having UX that helped with entering this data would be improving the quality considerably, such as CEDAR.
Carole Goble then talked about moving towards using Schema.org for FAIRer Life Sciences resources and defining a Schema.org profile that make datasets easier to use. The challenges in the field have been mostly social - there was a lot of confidence that we know how to solve the technical issues, but the social ones provide to be challenging. Carol named four of those explicitly:
- building consensus (it's harder than you think)
- the Schema.org Catch-22 (Schema.org won't take it if there is no usage, but people won't use it until it is in Schema.org)
- dedicated resources (people think you can do the social stuff in your spare time, but you can't)
- Build an ecosystem first, be technically light-weight (a great lesson which was also true for Wikipedia and Wikidata)
- Use open, non-proprietary, standard solutions, don't ask people to build it just for Google (so in this case, use Schema.org for describing datasets)
- bootstrapping requires influencers (i.e. important players in the field, that need explicit outreach) and incentives (to increase numbers)
- semantics and the KG are critical ingredients (for quality assurance, to get the data in quickly, etc.)
At the same time, Natasha also reiterated one of Mark's points: no matter how simple the system is, people will get it wrong. The number of ways a date field can be written wrong is astounding. And often it is easier to make the ingester more accepting than try to get people to correct their metadata.
Chris Gorgolewski followed with a session on increasing findability for datasets, basically a session on SEO for dataset search: add generic descriptions, because people who need to find your dataset probably don't know your dataset and the exact terms (or they would already use it). Ensure people coming to your landing site have a pleasant experience. And the description is markup, so you can even use images.
I particularly enjoyed a trio of paper presentations by Daniel Garijo, Maria Stoica, Basel Shbita and Binh Vu. Daniel spoke about OntoSoft, an ontology to describe software workflows in sufficient detail to allow executing them, and also to create input and output definitions, describe the execution environment, etc. Close to those in- and output definition we find Maria's work on an ontology of variables. Maria presented a lot of work to identify the meaning of variables, based on linguistic, semantic, and ontological reasoning. Basel and Binh talked about understanding data catalogs deepers, being able to go deeper into the tables and understand the actual content in them. If one would connect the results of these three papers, one could potentially see how data from published tables and datasets could become alive and answer questions almost out of the box: extracting knowledge from tables, understanding their roles with regards to the input variables, and how to execute the scientific workflows.
Sure, science fiction, and the question is how well would each of the methods work, and how well would they work in concert, but hey, it's a workshop. It's meant for crazy ideas.
Ibrahim Burak Ozyurt presented an approach towards question answering in the bio-domain using Deep Learning, including Glove and BERT and all the other state of the art work. And it's all on Github! Go try it out.
The day closed with a panel with Mark Musen, Natasha Noy, and me, moderated by Yolanda Gil, discussing what we learned today. It quickly centered on the question how to ensure that people publishing datasets get appropriate credit. For most researchers, and particularly for universities, paper publications and impact factors are the main metric to evaluate researchers. So how do we ensure that people creating datasets (and I might add, tools, workflows, and social consensus) receive the fair share of credit?
Thanks to Yolanda Gil and Andrew Su for organizing the workshop! It was an exhausting, but lovely experience, and it is great to see the interest in this field.
When I was a teenager I was far too much fascinated by the Illuminati. Much less about the actual historical order, and more about the memetic complex, the trilogy by Shea and Wilson, the card game by Steve Jackson, the secret society and esoteric knowledge, the Templar Story, Holy Blood of Jesus, the rule of 5, the secret of 23, all the literature and offsprings, etc etc...
Eventually I went to actual order meetings of the Rosicrucians, and learned about some of their "secret" teachings, and also read Eco's Foucault's Pendulum. That, and access to the Web and eventually Wikipedia, helped to "cure" me from this stuff: Wikipedia allowed me to put a lot of the bits and pieces into context, and the (fascinating) stories that people like Shea & Wilson or von Däniken or Baigent, Leigh & Lincoln tell, start falling apart. Eco's novel, by deconstructing the idea, helps to overcome it.
He probably doesn't remember it anymore, but it was Thomas Römer who, many years ago, told me that the trick of these authors is to tell ten implausible, but verifiable facts, and tie them together with one highly plausible, but made-up fact. The appeal of their stories is that all of it seems to check out (because back then it was hard to fact check stuff, so you would use your time to check the most implausible stuff).
I still understand the allure of these stories, and love to indulge in them from time to time. But it was the Web, and it was learning about knowledge representation, that clarified the view on the underlying facts, and when I tried to apply the methods I was learning to it, it fell apart quickly.
So it is rather fascinating to see that one of the largest and earliest applications of Wikibase, the software we developed for Wikidata, turned out to be actual bona fide historians (not the conspiracy theorists) using it to work on the Illuminati, to catalog the letters they sent to each other, to visualize the flow of information through the order, etc. Thanks to Olaf Simons for heading this project, and for this write up of their current state.
It's amusing to see things go round and round and realize that, indeed, everything is connected.
Over the last few years, more and more research teams all around the world have started to use Wikidata. Wikidata is becoming a fundamental resource. That is also true for research at Google. One advantage of using Wikidata as a research resource is that it is available to everyone. Results can be reproduced and validated externally. Yay!
I had used my 20% time to support such teams. The requests became more frequent, and now I am moving to a new role in Google Research, akin to a Wikimedian in Residence: my role is to promote understanding of the Wikimedia projects within Google, work with Googlers to share more resources with the Wikimedia communities, and to facilitate the improvement of Wikimedia content by the Wikimedia communities, all with a strong focus on Wikidata.
One deeply satisfying thing for me is that the goals of my new role and the goals of the communities are so well aligned: it is really about improving the coverage and quality of the content, and about pushing the projects closer towards letting everyone share in the sum of all knowledge.
Expect to see more from me again - there are already a number of fun ideas in the pipeline, and I am looking forward to see them get out of the gates! I am looking forward to hearing your ideas and suggestions, and to continue contributing to the Wikimedia goals.
Mark Stoneward accepted the invitation immediately. Then it took two weeks for his lawyers at the Football Association to check the contracts and non-disclosure agreements prepared by the AI research company. Stoneward arrived at the glass and steel building in London downtown. He signed in at a fully automated kiosk, and was then accompanied by a friendly security guard to the office of the CEO.
Denise Mirza and Stoneward had met at social events, but never had time to talk for a longer time. “Congratulations on the results of the World Cup!” Stoneward nodded, “Thank you.”
“You have performed better than most of our models have predicted. This was particularly due to your willingness to make strategic choices, where other associations would simply have told their players to do their best. I am very impressed.” She looked at Stoneward, trying to read his face.
Stoneward’s face didn’t move. He didn’t want to give away how much was planned, how much was luck. He knew these things travel fast, and every little bit he could keep secret gave his team an edge. Mirza smiled. She recognised that poker face. “We know how to develop a computer system that could help you with even better strategic decisions.”
Stoneward tried to keep his face unmoved, but his body turned to Mirza and his arms opened a bit wider. Mirza knew that he was interested.
“If our models are correct, we can develop an Artificial Intelligence that could help you discuss your plans, help you with making the right strategic decisions, and play through different scenarios. Such AIs are already used in board rooms, in medicine, to create new recipes for top restaurants, or training chess players.”
“What about the other teams?”
“Well, we were hoping to keep this exclusive for two or four years, to test and refine the methodology. We are not in a hurry. Our models give us an overwhelming probability to win both the European Championship and the World Cup in case you follow our advice.”
“For the European Championship?”
“No. To win both.”
Stoneward gasped. “That is… hard to believe.”
The CEO laughed. “It is good that you are sceptical. I also doubted these probabilities, but I had two teams double-check.”
“What is that advice?”
She shrugged. “I don’t know yet. We need to develop the AI first. But I wanted to be sure you are actually interested before we invest in it.”
“You already know how effective the system will be without even having developed it yet?”
She smiled. “Our own decision process is being guided by a similar AI. There are so many things we could be doing. So many possible things to work on and revolutionise. We have to decide how to spend our resources and our time wisely.”
“And you’d rather spend your time on football than on… I don’t know, healing cancer or making a product that makes tons of money?”
“Healing cancer is difficult and will take a long time. Regarding money… the biggest impediment to speeding up the impact of our work is currently not a lack of resources, but a lack of public and political goodwill. People are worried about what our technology can do, and parliament and the European Union are eager to throw more and more regulations at us. What we need is something that will make every voter in England fall in love with us. That will open up the room for us to move more freely.”
Stoneward smiled. “Winning the World Cup.”
She smiled. “Winning the World Cup.”
Three months later…
“So, how will this work? Do I, uhm, type something in a computer, or do we have to run some program and I enter possible players we are considering to select?”
Mirza laughed. “No, nothing that primitive. The AI already knows all of your players. In fact, it knows all professional players in the world. It has watched and analyzed every second of TV screening of any game around the world, every relevant online video, and everything written in local newspapers.”
Stoneward nodded. That sounded promising.
“Here comes a little complication, though. We have a protocol for using our AIs. The protocols are overcautious. Our AIs are still far away from human intelligence, but our Ethics and Safety boards insisted on implementing these protocols whenever we use some of the near-human intelligence systems. It is completely overblown, but we are basically preparing ourselves for the time we have actually intelligent systems, maybe even superhuman intelligent systems.”
“I am afraid I don’t understand.”
“Basically, instead of talking to the AI directly, we talk with them through an operator, or medium.”
“Talk to them? You simply talk with the AI? Like with Siri?”
Mirza scoffed. “Siri is just a set of hard coded scripts and triggers.”
Stoneward didn’t seem impressed by the rant.
“The medium talks with the AI, tries its best to understand it, and then relays the AI’s advice to us. The protocol is strict about not letting the AI interact with decision makers directly.”
“Ah, as said, it is just being overly cautious. The protocol is in place in case we ever develop a superhuman intelligence, in which case we want to ensure that the AI doesn’t have too much influence on actual decision makers. The fear is that a superhuman AI could possibly unduly influence the decision maker. But with the medium in between, we have a filter, a normal intelligence, so it won’t be able to invert the relationship between adviser and decision maker.”
Stoneward blinked. “Pardon me, but I didn’t entirely follow what you — ”
“It’s just a Science Fiction scenario, but in case the AI tries to gain control, the fear is that a superhuman intelligence could basically turn you into a mindless muppet. By putting a medium in between, well, even if the medium becomes enslaved, the medium can only use their own intelligence against you. And that will fail.”
The director took a sip of water, and was pondering what he just heard for a few moments. Denise Mirza was burning with frustration. Sometimes she forgets how it is to deal with people this slow. And this guy had more balls banged against his skull than is healthy, which isn’t expected to speed his brain up. After what felt like half an eternity, he nodded.
“Are you ready for me to call the medium in?”
She tapped her phone.
“Wait, does this mean that these mediums are slaves to your AI?”
She rolled her eyes. “Let us not discuss this in front of the medium, but I can assure you that our systems have not yet reached the level to convince a four year old to give up a lollipop, never mind a grown up person to do anything. We can discuss this more afterwards. Oh, there he is!”
Stoneward looked up surprised.
It was an old acquaintance, Nigel Ramsay. Ramsay used to manage some smaller teams in Lancashire, where Stoneward grew up. Ramsay was more known for his passion than for his talents.
“I am surprised to see you here”
The medium smiled. “It was a great offer, and when I learned what we are aiming for, I was positively thrilled. If this works we are going to make history!”
They sat down. “So, what does the system recommend?”
“Well, it recommends to increase the pressure on the government for a second referendum on Brexit.”
Stoneward stared at Ramsay, stunned. “Pardon me?”
“It is quite clear that the Prime Minister is intentionally sabotaging any reasonable solution for Brexit, but is too afraid to yet call a second referendum. She has been a double agent for the remainers the whole time. Once it is clear how much of a disaster leaving the European Union would be, we should call for a second referendum, reversing the result of the first.”
“I… I am not sure I follow… I thought we are talking football?”
“Oh, but yes! We most certainly are. Being part of an invigorated European Union after Brexit gets cancelled, we should strongly support a stronger Union, even the founding of a proper state.”
Stoneward looked at Ramsay with exasperation. Mirza motioned with her hands, asking for patience.
“Then, when the national football associations merge, this will pave the way for a single, unified European team.”
“The associations… merge?”
“Yes, an EU-wide all stars team. Just imagine that. Also, most of the serious competition would already be wiped out. No German team, no French team, just one European team and — “
“This is ridiculous! Reversing Brexit? Just to get a single European team? Even if we did, a unified European team might kill any interest in international football.”
“Yeah, that is likely true, but our winning chances would go through the roof!”
“But even then, 96% winning chances?”
“Oh, yeah, I asked the same. So, that’s not all. We also need to cause a war between Argentina and Brazil, in order to get them disqualified. There are a number of ways to get to this — ”
“Stop! Stop right there.” Stoneward looked shocked, his hands raised like a goalie waiting for the penalty kick. “Look, this is ridiculous. We will not stop Brexit or cause a war between two countries just to win a game.”
The medium looked at Stoneward in surprise. “To ‘just’ win a game?” His eyes wandered to Mirza in support. “I thought this was the sole reason for our existence. What does he mean, ‘just’ win a game? He is a bloody director of the FA, and he doesn’t care to win?”
“Maybe we should listen to some of the other suggestions?”, the CEO asked, trying to soothe the tension in the room.
Stoneward was visibly agitated, but after a few moments, he nodded. “Please continue.”
“So even if we don’t merge the European associations due to Brexit, we should at least merge the English, Scottish, Welsh, and Northern Irish associations in — ”
“No, no, NO! Enough of this association merging nonsense. What else do you have?”
“Well, without mergers, and wars, we’re down to 44% probability to win both the European and World Cup within the next twenty years.” The medium sounded defeated.
“That’s OK, I’ll take that. Tell me more.” Stoneward has known that the probabilities given before were too good to be true. It was still a disappointment.
“England has some of the best schools in the world. We should use this asset to lure young talent to England, offer them scholarships in Oxford, in Cambridge.”
“But they wouldn’t be English? They can’t play for England.”
“We would need to make the path to citizenship easier for them, immigration laws should be more integrative for top talent. We need to give them the opportunity to become subjects of the Queen before they play their first international. And then offer them to play for England. There is so much talent out there, and if we can get them while they’re young, we could prep up our squad in just a few years.”
“Scholarships for Oxford? How much would that even cost?”
“20, 25 thousand per year and student? We can pay a hundred scholarships and it wouldn’t even show up in our budget.”
“We are cutting budgets left and right!”
“Since we’re not stopping Brexit, why not dip into those 350 million pounds per week that we will save.”
“That was a lie!”
“I was joking.”
“Well, the scholarship thing wasn’t bad. What else is on the table?”
“One idea was to hack the video stream and bribe the referee, and then we can safely gaslight everyone.”
“We could poison the other teams.”
“Just stop it.”
“Or give them substances that would mess up their drug tests.”
“Why not getting FIFA to change the rules so we always win?”
“Oh, we considered it, but given the existing corruption inside FIFA it seems that would be difficult to outbid.”
Stonward sighed. “Now I was joking.”
“One suggestion is to create a permanent national team, and have them play in the national league. So they would be constantly competing, playing with each other, be better used to each other. A proper team.”
“How would we even pay for the players?”
“It would be an honor to play for the national team. Also, it could be a new rule to require the best players to play in the national team.”
“I think we are done here. These suggestions were… rather interesting. But I think they were mostly unactionable.” He started standing up.
Mirza looked desperately from one to the other. This meeting did not go as she had intended. “I think we can acknowledge the breadth of the creative proposals that have been on the table today, and enjoy a tea before you leave?”, she said, forcing a smile.
Stoneward nodded politely. “We sure can appreciate the creativity.”
“Now imagine this creativity turned into strategies in the pitch. Tactical moves. Variations to set pieces.”, the medium started, his voice slightly shifting.
“Yes, well, that would certainly be more interesting than most of the suggestions so far.”
“Wouldn’t it? And not only that, but if we could talk to the players. If we could expand their own creativity. Their own willpower. Their focus. Their energy to power through, not to give up.”
“If you’re suggesting to give them drugs, I am out.”
Ramsay laughed. “No, not drugs. But a helmet that emits electromagnetic waves and allows the brain muscles to work in more interesting ways.”
Stoneward looked over to the CEO. “Is that a possibility?”
Mirza looked uncomfortable, but tried to hide it. “Yes, yes, it is. We had tested it a few times, and the results were quite astonishing. It is just not what I would have expected as a proposal.”
“Why? Anything wrong with that?”
“Well, we use it for our top engineers, to help them focus when developing and designing solutions. The results are nothing short of marvelous. It is just, I didn’t think football would benefit that much from improved focus.”
Stoneward chuckled, as he sat down again. “Yes, many people underestimate the role of a creative mind in the game. I think I would now like a tea.” He looked to Ramsay. “Tell me more.”
The medium smiled. The system will be satisfied with the outcome.
(Originally published July 28, 2018 on Medium)
Today at work I learned about Saturn the alligator. Born to humble origins in 1936 in Mississippi, he moved to Berlin where he became acquainted with Hitler. After the bombing of the Berlin Zoo he wandered through the streets. British troops found him, gave him to the Soviets, where against all odds he survived a number of near death situations - among others he refused to eat for a year - and still lives today, in an enclosure sponsored by Lacoste.
I also went to Wikidata to improve the entry on Saturn. For that I needed to find the right property to express the connection between Saturn, and the Moscow Zoo, where he is held.
The following SPARQL query was helpful: https://w.wiki/7ga
It tells you which properties connect animals with zoos how often - and in the Query Helper UI it should be easy to change either types to figure out good candidates for the property you are looking for.