MapJakarta - Enabling civic co-management through GeoSocial Intelligence
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
| Title |
| |
| Title of Series | ||
| Number of Parts | 188 | |
| Author | ||
| License | CC Attribution 3.0 Germany: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. | |
| Identifiers | 10.5446/31593 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language | ||
| Producer | ||
| Production Year | 2014 | |
| Production Place | Portland, Oregon, United States of America |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
| |
| Keywords |
00:00
Projective planeDenial-of-service attackAreaResultantBitPower (physics)SoftwareWater vaporLevel (video gaming)Uniform resource locatorContext awarenessJava appletSoftware developerEngineering drawing
01:19
Water vaporContext awarenessData structureData managementPotenz <Mathematik>ResultantFlow separationSystem programmingMathematicsDependent and independent variables
02:01
Water vaporDisk read-and-write headBuildingMultiplication signBridging (networking)Denial-of-service attackData conversionPower (physics)Right angleSeries (mathematics)Context awarenessBit rateDependent and independent variablesDigital photographyResultantDataflowGroup actionData structureAbstractionChannel capacityFamilyLogic gateComputer animation
04:01
Multiplication signLevel (video gaming)Scaling (geometry)WindowNeighbourhood (graph theory)QuantumInformationFrequencyDenial-of-service attackDependent and independent variablesCurvatureAreaReading (process)
04:43
TwitterTable (information)Presentation of a groupTraffic reportingBitSmartphoneRevision controlSystem programmingData structureEmailPopulation densityLink (knot theory)Slide ruleLevel (video gaming)Denial-of-service attackInformationVideoconferencingNumerical analysisAreaComponent-based software engineeringInformation systemsMessage passingResponse time (technology)Open sourceSoftwareOffice suiteLocal ringVolume (thermodynamics)Uniform resource locatorDependent and independent variablesSpeicherbereinigungCondition numberOnline helpFamilyWater vaporGroup actionBridging (networking)Point (geometry)Hazard (2005 film)Bus (computing)Limit (category theory)Mobile WebDistanceProxy server2 (number)WordServer (computing)BlogData conversionMultiplication signReal-time operating systemKernel (computing)FrequencyEvent horizonProjective planeSheaf (mathematics)Service (economics)Workstation <Musikinstrument>Machine visionDifferent (Kate Ryan album)Process (computing)CASE <Informatik>FreewareBoundary value problemComputer animation
10:30
TwitterDistribution (mathematics)CodeFile archiverDiagramTraffic reportingForcing (mathematics)Term (mathematics)Computer programmingSystem programmingMusical ensembleDenial-of-service attackDependent and independent variablesProjective planeNeighbourhood (graph theory)AlgorithmServer (computing)Cross-correlationDecision support systemGene clusterPurchasingSquare numberLevel (video gaming)Context awarenessKey (cryptography)WebsiteBlock (periodic table)Uniform resource locatorTime seriesLine (geometry)Cartesian coordinate systemSuite (music)Multiplication signPoint (geometry)InformationTelecommunicationScaling (geometry)Event horizonProcess (computing)Front and back endsNumerical analysisGraphic designAreaTouchscreenFormal grammarDigital mediaDatabaseInformation systemsMathematical analysisSoftwareArchaeological field surveyOpen sourceService (economics)Software developerLink (knot theory)ChainNetwork topologyNumbering schemeWater vaporComputer architectureMereologyLogic gateSeries (mathematics)BuildingTotal S.A.Intrusion detection systemGame controllerGraph (mathematics)
16:16
Wave packetLine (geometry)Process (computing)RoboticsDenial-of-service attackMultiplication signDependent and independent variablesUniverse (mathematics)TwitterCrash (computing)Total S.A.Cellular automatonElectronic mailing listSystem programmingInformationInsertion lossSoftwareWater vaporSpline (mathematics)Self-organizationVector spaceNeighbourhood (graph theory)Level (video gaming)Scaling (geometry)ComputerDiagramResultantCASE <Informatik>BuildingMashup <Internet>Clique-widthCodeDigital library2 (number)Traffic reportingEvent horizonGame controllerKey (cryptography)Open sourceLecture/ConferencePanel painting
21:25
FreewareDependent and independent variablesForm (programming)Service (economics)Data storage deviceSoftwareSlide ruleOpen sourceDenial-of-service attackMathematical analysisUtility softwareBuildingLine (geometry)Computer animation
22:17
CuboidWordDenial-of-service attackWebsiteDisk read-and-write headMessage passingOffice suiteSocial classSoftware developerProcess (computing)Volume (thermodynamics)Traffic reportingStudent's t-testCombinational logicFrequencyOpen setGraph (mathematics)Natural languageLimit (category theory)Computer programmingCompilerKey (cryptography)Data conversionPressureTwitterAddress spaceFilter <Stochastik>Wave packetMultiplication signWindow
Transcript: English(auto-generated)
00:00
applied research like this project. So to provide a bit of context, Jakarta is home to 27 million people, so the population of Australia in one city, and experiences seasonal flooding as a result of the monsoon season. So to give you a bit of a background, this map shows the location of Jakarta
00:21
on the island of Jakarta in the archipelago of 13,000 islands that makes our country of Indonesia the fourth largest country in the world, just to the north of Australia, largest by population at least. So to understand the concept of flooding in Jakarta, we need to go back to the original settlement,
00:42
the settlement of Batavia in the 1700s. So whilst the other colonial powers may be quite happily sailing past this low-lying marshy area on this tropical island, the Dutch spotted this as a potential area to employ their methods of drainage and clearing using a canal and a pump network to provide a city which
01:03
was ideally located for trade. So water has always been at the heart of the situation of the city of Jakarta. There are 13 rivers that flow through the city from the mountains in the south to the sea in the north, and the land is at sea level.
01:20
So to put that in the context of 27 million people and put that in the context of the 21st century, Jakarta is the world's most urbanizing, rapidly urbanizing place. These astromeges show that change, that almost exponential growth of population and also land use change.
01:40
And so what was potentially a sustainable system that the Indonesians inherited from the Dutch after independence in 1945 is becoming now increasingly unsustainable because there is too much water in the system, and the management of the infrastructure results in severe fling. So the infrastructure kind of looks like this.
02:00
So you've got 13 rivers that flow through 27 million people. And so all of that water has to be pumped over the seawall or pumped through a series of canal locks and gates. And so if we see one of the pumps here, the pump on the pump station on the left was the one that exploded during the 2007 floods.
02:21
It was working at 30% capacity. It needed to be working at 120% capacity to cope with the deluge of water. We see rapid urbanization, buildings like this, which because of groundwater abstraction are resulting in a subsidence rate of a quarter of a meter per year. So when we look at the pumping of all of that water from those 13 rivers over the seawall into the ocean,
02:41
you'll notice that these new pipes are tethered to that concrete block, which is being subsided by the groundwater abstraction of the nearby urbanization. Moving on to the prospect of climate change within Jakarta, the guy on the right-hand side eating his lunch there, you see that he sat on a bridge. That's the 2007 flood mark above his head.
03:02
So in 2007, there was a large-scale flood. Up to 40% of the city over the monsoon season was flooded. The government swung into action, moved people out. The flood receded, and people moved back to clean up and to restore their homes. The photo on the left was taken by one of my colleagues at the end of the monsoon season this year,
03:20
and you see the flood mark above the guy who's cleaning outside his house. Because of climate change, there was 50% more water in the city of Jakarta this year. And so this flooding happened six times. This guy had to move his family out, try and keep them safe, go back, clean, tidy up, restore, go back to work again,
03:41
and then two weeks later, he was on the move again because the city cannot cope with the response of water as a result of climate change. So thinking about urban infrastructure resilience in particular a geospatial context, what does that look like, and how do people interact with infrastructure both during the dry season and during the wet season?
04:02
So the government response kind of looks like this. Have an open geostat predominantly using quantum GIS to produce static maps over a six-hour period like this one, which form the response of the government to flooding. So which of these neighborhoods are flooded at any one time dictates where resources need to go.
04:21
So clearly, when you're thinking about flooding over that scale of people and area, six-hour time window is quite a long time to be able to formulate or quite a long time to formulate your response. So anything that we can do to decrease that time to speed up the response to gather more information
04:41
will be of benefit. One of the interesting things about Indonesia is that it has the highest density of smartphone and cell phone use of anywhere in the world, 160% penetration of mobile phones. So more than 50% of the population have two phones.
05:01
The highest number of Twitter users anywhere in the world in Jakarta. Twitter has opened an office this month because they kind of realize that it's of interest to them in this area. So there's a question. All of those people are on the ground as a big data source. But what we need is not big data, we need smart data.
05:20
So can you argue that people with smartphones, which are geolocated by GPS, are the potentially smartest sensor that you have to tell you about the situation on the ground? And can you harvest that information in some meaningful way to then tell you about the response of the city to an event? Just before I show you a very short video
05:41
explaining more about the project, this slide's great because it shows an informal economy in Jakarta of 60% of the economy is informal. But these guys who are garbage collectors have swung into action and are making a quick book by transporting this guy across this flooded intersection of road. Whilst he is simultaneously tweeting out
06:01
the location of that crossing point to tell all of his friends of that's how you can get to work. So we have developed a system that collects tweets, collects geolocated tweets, puts them on a map for a public information service, and also aggregates that data, and we built a data API for the government to use as well
06:21
so that they can try to improve their response times. So this video is what we plan to tweet out to all of the people in Jakarta that talk about flooding this year. So when you're having a conversation with your friends on Twitter, and you mentioned one of the keywords like the word flood, the word banji, Bahasa Indonesian, you'll get a tweet from us automatically
06:42
with a link to this video. There is a new tool in Jakarta bringing together mobile mapping and local flood information. This community flood map is available anywhere,
07:02
alerting you to water impasses in real time to help you navigate the city. We know that the citizens of Jakarta have the best information on flooding conditions. You are already tweeting each other, helping friends and family avoid hazards around the city. PETA Jakarta uses this on the ground information to give you a comprehensive map of the flood conditions.
07:23
When you see a flood, tweet banji at PETA Jakarta and your report will appear on the map, alerting the community to the flood. Remember to turn on your phone's geolocation so we can pinpoint the report. The more people use PETA Jakarta, the better the map will be. Working together, we can help everyone
07:40
bypass flooded areas, saving time and avoiding danger. Visit PETA Jakarta.org to get started. So, what we've done, and I'll talk more about the technology in a couple of slides time, what we do is we connect to the Twitter API, we listen for those keywords, we communicate back to the communities
08:00
and the individuals asking them to confirm the report flooding, similar to the Ushahidi model, which was mentioned in the last presentation of having a confirmed and unconfirmed report structure. And so then we produce a map, which I'll show in a second, of tweets across the city in real time relevant to flooding. When we tested this this year,
08:20
in one 24 hour period, we had over 150,000 tweets that were all about flood, of which about 3% were geolocated. And so there's a large volume of data, and what we did is we thought we'd invented this amazing system where you're talking about flooding and we send you the least creepiest message you've ever imagined.
08:41
It's like, hey, we heard from a guy who's my friend that maybe flooding was like, as opposed to we're watching you, like we're the CIA or whoever, and you know, where's the flooding thing? So we came up with this idea, and saying, oh, like we heard that you're talking about flooding, like confirm it, tell us, like tweet at us, put geolocation on. And essentially what we'd invented there was spam.
09:02
And so within that 24 hour period, there's a limit which Twitter imposes on you on how many spam messages you can send, which is 250, and we attempted to send 75,000 original users messages in 60 seconds. Now all of a sudden I was reading the server logs and everything just stopped.
09:20
My colleagues sat across the table from me in Jakarta and went, I just got an email from Twitter. And so it was an interesting experience, and now we're working with Twitter to make that process enable and follow the guidelines and for them to help us with that. So the public information system for the individuals in Jakarta looks a little bit like this,
09:41
and this is actually a bit of an outdated screenshot from our previous version. So when you're in the city on your smartphone and you have geolocation enabled, you get zoomed in to where you are on the map to show you the reports, both confirmed and unconfirmed around you, and to show you the hydrological infrastructure network as well. So that's the public information component.
10:01
So it's like help yourselves, help them up, crowdsource, what's going on. We'll see a quick example of that. This was captured this year. So I walked down this street in the morning and it was dry, there was heavy rainfall over the lunchtime period, and then it was raining. And this woman tweeted, and she's trying to get to the bus station, which is the concrete bridge you can see in the distance.
10:21
She's got geolocation on. So if you then look at the map with a device, in this case, my map, you can see that report geolocated and pinpointed on the map. That's pretty cool, because what that gives us is a public information system where people can crowdsource situational reports. Many people in Jakarta already talk about flooding
10:41
using SMS chains and Twitter to inform each other of what's happening in lieu of the official alert service from the government. So then the question becomes is, can we simultaneously provide that system, but can we also provide a geographical information system for the government for them to make more informed decisions about the response of infrastructure to flooding?
11:02
Because remember that the flooding isn't traditional inundation. The flooding occurs because the canals are full and part of the infrastructure fails because it cannot cope. The pump explodes, the gate gets left open, the wall collapses, resulting in inundation of the city. And so the response of flooding is an infrastructure challenge as opposed
11:21
to minimizing the amount of water. So managing that infrastructure and working out where to maintain that infrastructure is of prime importance. One of the things we've been very successful with in this project is our application to the Twitter data grants program. And so we now have access to a historical archive of tweets to test some of these ideas.
11:41
So if we see here the monsoon season from 2012, 2013, the spatial distribution or the apparent distribution haven't yet run a clustering algorithm on this appears to be phenomenal across the city. One of the things we see in the Western world is many examples of people looking at Twitter and then looking at the device people use to tweet.
12:01
For example, in New York, where are they with iPhones? Where are the people with other devices? And they showed that there was a correlation between that differential and social status. One of the really interesting things about Jakarta and developing nation is that we don't see that. We see an almost even spread of tweets across the city from both poor and rich.
12:21
So then if we go to look at a time series of can we use all of those tweets in a big data context as a trigometric to say flooding is happening at this location at this point in time. So this graph shows the 2013, 2014 monsoon. Some data from January, 2014, the blue bars represent
12:41
the total area, the total number of neighborhoods flooded at any one time in the city whilst the green line represents the Twitter activity relevant to flooding. So it would appear that there is a relationship between Twitter activity occurring and flood events occurring in the city.
13:01
So what we're doing on that same map that you saw before and that same website, when you zoom out of all those individual tweets, you can get an aggregate report like this which is at the same spatial scale as the government work app. So this year moving forward, we have a joint pilot study with the flood management team in Jakarta and they'll have two maps on their screens
13:22
in their control room. They'll have our map showing Twitter activity and then they'll have their map from their traditional formal sources of flood information and they'll be able to compare and contrast to see what the potential benefits of using something like this are. Can they identify floods sooner and quicker and earlier or are there problems with using this kind of data?
13:43
And the other thing that we've been doing which is quite novel in this context is gathering a full suite, a spatial topological suite of the hydrological infrastructure system. So rather than just having all of the spatial data about where the rivers and the pumps and the canals are we're building a spatial topological model
14:01
so that you can say what is that pump connected to and then infer how many people will be affected if that pump broke or downstream was affected of that particular link. So there's quite some interesting development there in terms of how we use postures at the backend and how we represent that topology as a series of nodes and edges, node IDs and edge IDs
14:21
within our posterior schema. So to kind of summarize what I've said already, this diagram is what happens when you give an architecture diagram which was quite simple or so I thought to a designer. Graphic designers seem to love to produce this kind of tube, London tube kind of looking map or it does to me but very simply you've got reports.
14:45
Let me see if this works. You've got some Twitter reports coming in here. You've got some field survey data coming in here of the hydrological information. We have some custom node.js code that we've written to collect and aggregate all of that data and push it into our postures database. We then pump that back out to our server
15:01
to produce both our public information system on Pareto Jakarta and to do our offline analysis to help the government make better decisions. And so that suite of software which we have collected which is all off the shelf, so a postures backend,
15:21
node.js software, RPO and GeoJSON to build our data API. Those are all Lego blocks of the existing geospatial community which is great which we packaged together in a suite called Cognicity. And Cognicity is really all about listening to social media, so collecting information, doing the filtering
15:40
and analysis that you need to do on that information and then communicating that information both to individuals in the communities and to the government and then enabling the communication between those two strands so that we can move forward and we can use all this spatial data as a process of civic co-management because essentially that's where we're aiming to go is that the people know what the situation
16:02
is on the ground. The government knows what they can do in terms of response. How do we put the two together? How do we use FOSS 4G to handle that big data problem to connect those two groups? But it's not just big data because one of the things that we stumbled across
16:21
in working with many of the community organizations in Jakarta is that they are already in many ways incredibly resilient to the process of flooding. And it's only really since the 1970s that flooding has been termed a disaster. So before that time, many of the older people in the community remember flooding as a time of plenty
16:41
because it flooded, the water was clean, it was pre-industrial, fish were just flowing down the river and you could scoop them out of the river and everyone ate fish for about three months because it was great and so everyone was swimming in the river and playing around. And so these communities have evolved over time to respond to flooding in their own way.
17:00
So this is Whittier, who is our amazing person on the ground in Jakarta and she's holding on to a rope line which the community in this informal settlement has put in themselves to drag themselves along during the flood because the water can be off to the first story. Everyone in the community knows where the rope lines are and the community pre-organizes the evacuation network of all of the elderly people within that slump.
17:23
So all of the old people are evacuated as the water rises along these rope lines into first and second story buildings. Now one of the problems is that when the government responds to the flood, and this isn't to say that we're backing the government's response because the enormity of the challenge is phenomenal,
17:41
but when the government responds one of the things that can happen is that the boats that they use respond, crash into the rope lines and the propeller cuts the rope line which is no longer there and so then the people who are at the end of the rope line cannot access the rope to pull themselves along to stop themselves getting sucked down the drains. So how can we produce a GIS system
18:03
using an open source software that allows those individuals in communities to map those rope lines to then communicate that information to the government? And one of the really interesting things is that many of these communities have already produced their own maps.
18:21
They're not geospatial maps, they're diagrams. And so a better geographer than I would be able to tell you some really interesting things about this map, such as the width of the street. I mean, you saw that this street is not very wide. This is the place that the community exists in. People talk, people shop, people communicate,
18:40
people work, people sell things. So conceptually, people think of these streets within their informal settlement on the edge of the river as being a wide avenue. It's not. And so when you try and de-erectify that using some control points, you have this horrendously fitted spine to try and make sense of that information
19:01
within the rest of the world. The key thing here is that the communities are resilient. We've got a small data approach to try and collect that information and mesh it with our existing system to provide a cohesive understanding of the response to things. So what we did is we developed a system. Again, you send us a tweet,
19:21
but this time we're training people in the community to tweet specific codes to us so that we can do pre-flood and post-flood damage assessments so we can understand what things are going on in the community, where the rope lines are, what things they're doing to get ready for the flooding, and then after flood, what was the result. So very quickly, within one day,
19:42
we can train our community leaders to go into these urban villages to understand the response of these informal settlements using Twitter. And because everyone else is on Twitter, everyone can follow and see where the data is. So whilst many of these community organizations already did this using a paper system, often that data just went in a drawer
20:01
and never got digitized. But now we took it straight from Twitter and we put it on a map so that people can see it instantly. Oh, yeah, that was my report. I did that. So here's an example. At MapJaccar, geospatial rapid assessment platform, house 52, five people in the house,
20:20
15 days of work, 1.5 million rupiah lost as a result of the flooding, 45 days of school collectively from five children, 30 days of sickness in total in the house, 220,000 rupiah of damage, and the flood was at shoulder height. Because no one works in meters or feet, people work in both parts.
20:41
What that allowed us to do in one day with five volunteers was to map for the first time the impact on an informal settlement in Jakarta of a flood event. So as we increase the flood heights along the x-axis, we see that this appears to be a proportional increase in income loss and damage by household.
21:03
Taking that one step further within our system, we then take those maps which we roughly geo-rectified and I just did this very quickly one afternoon, but we can then attribute which houses were hardest hit and which house is a first and second story and where those rope lines need to be
21:21
to produce a neighborhood scale map for the government so that the communities can work with the government to tell them where the rope lines are and what they're doing to provide a more organized response. So to summarize that whole thing I guess in one slide,
21:40
we have a big data approach. We're capitalizing on the utility of free and open source geospatial software and services to do big data geo-analysis and big data geo-storage and build our platform, but we're also integrating small data using very similar tools to do targeted data collection to try and improve and build a holistic understanding
22:03
of the response of the city and how we can improve that response to flooding in the city of Jakarta. One last slide, thanks to our funders who have been very generous and that's it, thank you.
22:27
Any questions?
22:41
There's a lot of potential there, yeah.
23:08
You don't need to. We're training the people and turn geo-coding on because like the keynote or the opening address this morning it's really hard to do geo-coding and it's really hard to do geo-coding when it's in a different language.
23:22
And so for us, maybe we've seen in the literature that you can do increase geo-coding in tweets by maybe 10% because you have only got 140 characters. So that's a lot of effort to expand for a very minimal increase in the data volume. So what we're doing is trying to encourage those people of like turn your geo-location on and what we do is if you send us a message,
23:43
a report, a confirmed report of flooding and geo-location isn't on, then we flick your tweet back saying thanks, but next time you just stick the phone out the window. So just put your GPS on to try and encourage people to do that. Did you say you're targeting the Twitter media?
24:07
Yes, we're using our own tools. So we built, we developed this program within a 10 month period. And so we did some very rapid development using Node.js because the data that we wanna collect and then we wanna push the website is very targeted
24:22
and it's all in TopoJSON. To answer the second point, when we tested using the streaming API, we didn't yet hit the rate because of the filters that we put on those keyword and the bounding box. But one of the things that the data grant has enabled us to do is to start the conversation with earlier about how this potentially would scale
24:42
because I think we will hit them. I mean, I think it'd be, if we hit the limit this year and that's a success because more people, more data than we can get. Yeah, so we did a workshop with a couple of students
25:00
for a day to try and then we tested on the fly the different keywords. So we have six words, which are three, the same three in English and Bahasa Indonesian. So it's off the top of my head, flood, pool and submerged and we tried quite a few others because there's some other keywords in Bahasa like rising and what we found with a few of them
25:20
because the students were getting really excited like, yeah, everyone uses the word rising, everyone uses this word. But then we were getting all sorts of like my blood pressure's rising and we're just getting all of these tweets all over the place. So we kind of had to slim it down. So it'll be interesting to see once we get the text in this year and then also from the data grant to see what those combinations of keywords are to try and work out any that we're missing.
25:43
Cool, thank you.