OEKG: The Open Event Knowledge Graph

Video in TIB AV-Portal: OEKG: The Open Event Knowledge Graph

Formal Metadata

OEKG: The Open Event Knowledge Graph
Title of Series
CC Attribution - NonCommercial - NoDerivatives 3.0 Germany:
You are free to use, copy, distribute and transmit the work or content in unchanged form for any legal and non-commercial purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Release Date
Production Year

Content Metadata

Subject Area
Accessing and understanding contemporary and historical events of global impact such as the US elections and the Olympic Games is a major prerequisite for cross-lingual event analytics that investigate event causes, perception and consequences across country borders. In this paper, we present the Open Event Knowledge Graph (OEKG), a multilingual, event-centric, temporal knowledge graph composed of seven different data sets from multiple application domains, including question answering, entity recommendation and named entity recognition.These data sets are all integrated through an easy-to-use and robust pipeline and by linking to the event-centric knowledge graph EventKG. We describe their common schema and demonstrate the use of the OEKG at the example of three use cases: type-specific image retrieval, hybrid question answering over knowledge graphs and news articles, as well as language-specific event recommendation. The OEKG and its query end-point are publicly available.
Keywords event knowledge graph, open event-centric data
Satellite Pattern recognition Suite (music) Greatest element Presentation of a group Context awareness Graph (mathematics) Weight Multiplication sign Numbering scheme Open set Icosahedron Mereology Tracing (software) Formal language Neuroinformatik Information retrieval Medical imaging Benchmark Different (Kate Ryan album) Hypermedia Query language Software framework Information Endliche Modelltheorie Extension (kinesiology) Social class Physical system Pattern recognition Link (knot theory) Theory of relativity Keyboard shortcut Bit Open set Formal language Demoscene Software framework Website Right angle Quicksort Procedural programming Freeware Resultant Directed graph Point (geometry) Trail Slide rule Link (knot theory) Observational study Set (mathematics) Connectivity (graph theory) Disintegration Computer-generated imagery Student's t-test Event horizon Computer icon Theory Time domain Frequency Latent heat Goodness of fit Causality Representation (politics) Computer multitasking Directed graph Graph theory Task (computing) Domain name Execution unit Stapeldatei Graph (mathematics) Information Uniqueness quantification Projective plane Interactive television Content (media) Set (mathematics) Extreme programming Mortality rate Parity (mathematics) Grass (card game) System call Subset Number Word Uniform resource locator Event horizon Personal digital assistant Hybrid computer Revision control Formal grammar Musical ensemble Game theory Library (computing) Singuläres Integral
things are also the paper saw year was in such as long as they're actually said made as its combined in the oem even modest growth and all of the status s. are some more related to the clover to project not been some or related they were actually.
created as part of this project from all these different the ph d. students mentioned soul you first saw them and dismiss this may be slightly different but there is even kg lied which a satellite working off event kg which was just briefly and mention by and then at the very beginning of this work. so which is our marketing which events centric much class get which cover slots of events in temperature data had kind of success a car all the order event no trossachs of which all later and yet we created a new light but who are close it to make he said users who was in a week and she then small so. for most of them are publishing other papers say which also as sided and my right in this case the paper s. all i mean i can't just say one sentence for them but of course you could stay much or about each of them saw the first one su and k.g.b.'s trick which is a data set of language specific events center. the user interaction traces which we can use for recommendations as i will also show a to an example to us weekly which is a question answering datasets which we can change but i special and answers yes the m.s.m. dataset to which he meant use for initiated because it's kind of the mighty more data set them useful might attack. cloning we have set intimations fretting over new status said which was just mentioned i think to presentations to go by a good. and we have one more or a new suite at a deficit quite time which is reduced the city has used to take us them at all to get to specific topics olympic legacy and you risk of diseases i'm so you can reduce its new subjects to have a detailed look at these two topics and at last we have to your at it. said which was also introduced at the beginning of this workshop gushed an which is a at us a new entity recognition framework so you see these are kind of many data sets from different domain so on and yet we integrate them to get that to create this. or even new for us and to do so or its first seven look at what it actually looks like saw the f. the year and example event in the oil kitty of course of just a except of everything we have failed to spot to an end point and saw but we also it this way to we look into the event saw us the excitement of the. but to fire and paris yet flavors and many and which just as crucial many languages and of course also many relationships between this event and other events at such events and so on. but at thanks to the status its be integrated we can go a step further and also add an extended was very very different educational men so he wanted it to see a we actually had a news article best part of the always kg which is about the olympics so you can really get a different perspective of of the olympic games that much. just that cities yes classic relations between the sea on the right we also have an example which just from the question answering data said which actually contents a question and answer to about the apollo eleven a submission saw and all the things are of course integrated into a column on the trot so. for it so that the question your resources specific apollo event and so on. in to do this i mean that talks shown to the seven data said some to question is how we can we put them to get us all goal was to transform this data sets of to sing a integrated knowledge trust that provides links between all the invoice resources rightly daughter at second estimated cross want to put them together. saw and we want to make this including of new data system to the opening event not to trust simplest possible us all seeing that yet seven data sets we wanted to fight an approach which makes easy also for the different ph d. students that ended their data sets and to it and so i will show you this. i'm excited about what we actually did this we start each day a set from course different data sets a shock to us with a specific name grab your eyes so if you're familiar with the idea with this concept you are that if you may know what works so you can assign. suspected extreme specific part of trouble switched on and on the track you can decide that a specific you are right so too easy to get to address and more easily on and what we did this we decided to take even katie light se yet or richness source for all these events and entities because of course you some on you to agree on a. common reprimand representational c. suite of words and concepts like events he thought what cop or entities are a bomb or something so we are actually using your ice taken from the event kg by data set to have a common m. us are set to open looks like this. so it actually like the scene from a technique that bind us you what everyone needed to do to add a data set to that we can she was too kind of call a m r two left to f.b.i. cause and to create the grass saw step by step such a short year maybe we can first look at this image on the bottom of a we had this. even kids the name graph best part of the only country in their already but now i'll assume your to it and your dentist said and at that you are i knew graph are and what you first need to do us a tool get this into a link between you and k.t. lied and your craft so you're actually use or we get data. and he's already be your eyes every tool you get this get the proper you are isolated away kg. which makes sense because we see that we have an updated news articles which we could be a concept and so on which we saw the ad impressions fighting all the music stopped but i think so that was always some connection given that say nine and handed having these us from a simple at the icon you can melt use it as senate. i get libraries at to create a grass and six to realize it into a fight it and then uploaded this was another baby i call and the identifiable from him cross into the only case she then of course it's not only data we all sorts of steam are so is possible. we wanted to or simply use the scheme which is already reduced by at the by the event kg died. if that's not possible or yet we tried to stand up look at your various that is essential possible b l work on your special session priestly my tent extensions to the boy kg soul for that we also use similar procedure so you could also update the scheme notes stigma dates. using an f.b.i. and the end that's what our city most like which i won't show up and i would die show it in every day but i don't talk about it in detail important history of on the bottom let us the white part which is kind of peace a part of thing it can be larry deal with the fence entities stop time off events and so on. and then we have these different data sets which were at also contribute to specific parts of the scheme for example at the top left and the green we have to question answering dataset we conduct which consume do we use steam out of our question i'm supposed to eat and then use that doing this to the it yet. to the existing scheme i also just one more thing to show you to make another reference to what we got into a sweat shop is feel set this news article datasets where we also a model of the segment of the sick so i'd be solved that settlements ours it's its specifically interesting topic aside the stomach. when you than less of a little bit twee some posts what can we actually do was the only event nonprofit so the first one is called that image which us all because yet we also ready sought the i'm presentational today about the. i think from good at it about claims and imagine us and some so we always see that the mortality is an important aspect year soul one other task could be given credit occasion image shows which this year often use the computer which will end for a sudden such vast you do beauty need to wear the five class hybrid. king and unique image just and those that are part of the only kg thanks to him and davis said and to your narrative to set new saw what we can do as if you try this was experience. you can actually at this theory for him a trip to the way we asked the question is moving images of locations where airstrikes segment on soviet and thinking this is something we can also have a shot look at the coffee we are where we are basically really asking for something tight as i just thank you think that you are tied hierarchy. on then we're asking for the place of this earthquake which actually using information from even kitchen by so i tested location of events on and find in the u c m m dataset which was just like model dataset tool as a retreat the images such patients. and if you read this question and a joke period ahead and two we get results like this but compete but some ice examples where insurgents the names of these patients were once there it could some us earthquakes and we get proper images of these reaches a second example this event recommendation saw. to which typically means and you want to get retreated just as relevant events in a particular context and we have health was saying on the context of languished this is indeed saul want to all retreat events which are related to us said. it will give vent that i'm from which specific q so i wish all the excitement and neck slide but we also do we enrich studies and information so that six ft using the two data sets stated here at the bottom on there is i just saw i was gives a spike could be a formal. now she just at its other in the question which events are related to the first word or from russia point of fuel cell maybe one or two point to sell the such for the a one identified but that the media your eye and research for this residents thought the russian credit score and food at it by. having a batch of presidents year and interesting is what we see of the bottom of the actually sites say we s. at st events which us such highly related from the russian point of fuel and reassessed up to its taken from you and pity fight. ride and that's exactly what i want to say able to eventually craft a want hence they could get in on those who are on dismiss who contributed to it and again that sink to the we saw so few free tool that that backside and have a look at it had also expose some are some plus which russia.