Notes: State of the Map 2015, NYC (Day 2)

State of the Geocoder

Harish Krishna, Mapzen @harizh

• nodes without names: many of them exist and they aren't terribly helpful. There isn't a real workaround, other than prompting the user at node creation time to "guess the park name"
• there are 7,000,000 + non-unique nodes in the database (shared geography). The solution here is to dedup the data by consolodating tags into one node. This is expensive but pretty necessary. Solution: make sure there are warnings at node creation time if geometry is shared
• consistency: misspelled tags, misused key/values, more than one way of expressing a feature. Solutions: add maproulette task, mapping parties, bots to fix typos in tags
• address tags add structure to street names, they are important for address parsing - these are the key counts for the addr tag - they should hypothetically all be matching but that's not realistic
• addr:interpolation is critical for address geocoding, but very low coverage at present

Rahul Maddimsetty, Foursquare @rahulpratapm

• What will it take to make OSM more competitive with proprietary mapping data?
• data quality/quantity comes with the developed world
• "anyone can build an 80% geocoder" - unambiguous, fully-qualified, boring queries. The last 20% takes a lot longer
• building features into engine: fuzzy matching, hacks around regional/special cases
• what makes a geocoder weep? distinct address ranges for each name, 3 x 2 city/postal code context, some of this is poorly modeled in TIGER import
• the ease of tagging in OSM is great, but it has implications on the geocoder - makes the map great for rendering and routing the visual world. Geocoding requires unseen address ranges, city/postal code contexts. The problem with the unseen features is that they are established by authority, which means it is all based on bureaucracy
• current import guidelines actively discourage anything that is non survey-able. Parcels are "not useful on a map" because they aren't visible, but this would help geocoding
• separate the core from meta data to make sure geocoders can fuzzy search
• he doesn't buy the argument "it's hard to get mappers to care about relations" - there are no returns on creating data for geocoding. We shouldn't even try to get the community to start caring about this, and instead write bots to start editing our data!
• geocoders have a tough time because they directly expose the underlying data

Britta Ricker, Education in a changing GIS landscape

• How did she get into the mapping world? She didn't know about it as a younger kid/adult. When she was deciding what to do she sought out a skill - not a specific one, but a trade. She wanted to do international development, but found mapping through an academic mentor.
• GIS is an exciting realm to be teaching in because the landscape is constantly changing. This means we will always need more people to help us map it.
• The technology in the mapping world is constantly changing. Now more and more people have access to these technologies
• The thing that remains constant: gathering an inventory of spatial information to reveal "unknowns" or patterns in the world and make decisions
• Another constant: the challenge of teaching GIS.
• We love OSM because of the invitation to participate in the process. Everyone has an opportunity to participate in the entire spectrum of spatial data.
• OSM can be used as a tool to learn about the GIS process, but also to involve them in the creation of data in the world, which make them understand the tool.

How do we develop curriculum to prepare students to get jobs?

• build on what the student already knows, but how do you do this with so many different backgrounds in the GIS world
• "I was afraid to get out of my GIS toolbox" and was intimidated to enter the programming world, but found that entering through her understanding of maps (changing a lat/lng pair in javascript) was helpful
• Users of GIS as a pyramid:
• base: Basic user: comfortable with using a map and getting information from it
• middle: Traditional GIS students: can make maps, gather data, do analyses *top: CyberGIS Experts: those with the programming knowledge to continue to build
• But maybe instead of a pyramid, we are a merry-go-round. We are all helping and contributing to a larger entity

Goal: attract new (and more) mappers!

• we need all of the different actors in the mapping world to make sure it keeps moving forward. But we aren't always sure that the others exist - we need to remember! - recognize that each person has a unique role in the mapping world
• if we don't recognize the entirety of the mapping world, the map becomes just a reflection of ourselves
• how can we get more mappers? Engage students, listen to them, understand what makes them excited, and point them into the right resources for continued education
• don't build a curriculum on strict, immovable examples and data - you need to be able to learn to structure a project on your own, find data on your own

Drishtie Patel, Missing Maps

• What is missing maps? Mapping the most vulnerable places in the world. Starting in areas where they already have projects through their partners (Red Cross, HOT
• Core ethics: open, respectful, building local capacity. Want to prioritize people over data
• hosting a bunch of mapathons (Map off!) between schools. It's a fun way of getting people involved
• so far they have hosted 46 mapathons in 11 countries
• ~ 3.7 million edits to OSM

Case study: Zimbabwe

• doing a long term integrated urban resilience program - started with baseline surveys to determine key areas where they can work.
• involved some training with GIS professionals - included theory and practicing what volunteers would do on the field. Focused on community discussion to gain common ground on differences between individuals in the area. They were able to determine what to map, and what spatial information was important to the community
• had to figure out how to work with temporary structures (even if they weren't that temporary), how to handle new tags
• hand-drawn maps to see what people knew about their area, how they perceived it
• "Mapmobile"! (think Batmobile)
• community relationships == excellent (and important for successful outcomes)
• many questions about how to tag things: pretty sure what secondary roads look like in the USA, but what about in Zim? They spent a day taking photos of what secondary roads were considered so they could show volunteers and get more accurate data
• use JSOM because wifi is a huge issue in these countries

Tanzania

• have since done some similar projects in Tanzania now, but they incorporated a "recap" of what volunteers have learned, what they grasped, and what they needed help with
• why was "building material" important? Found that those in permanent structures were getting healthier than those not in permanent materials so they were able to prioritize where to help
• road mapping is incredibly important because of the different rain seasons

Lessons Learned from international community mapping

• community relationships are the most important
• able to do mapillary
• be flexible - things happen quickly and don't happen as expected
• have fun - it can be challenging and stressful, education is important. It's about inspiring people to add quality data
• respect tea time: take breaks and get into the culture
• don't expect power or wifi (even from hotspots)

Courtney Clark, OSM at Peace Corps

@courtneymclark3

• OSM aligns well with the gov support of the open initiative

They are working on three big opportunities

1. local OSM capacity building

• OSM toolkits and community mapping training guide
• OSM has transformed the peace corps experience: a volunteer learned about OSM, went to her host city in Botswanna, and crowdsourced the entire city data and added to OSM. Host community added local knowledge and volunteer & health clinic conducted geocoded survey after an indoor residual spraying campaign to reduce malaria. They created a map that shows which houses were sprayed to kill mosquitoes, and which were not. Because of OSM, making spatial decisions was possible.

2. crowdsourced mapping by PC alumni

• Crowdsourced community mapping led by over 200,000 volunteers
• Here is an OSM dashboard of all peace corps volunteers

3. crowdsourced mapping by America students

• lots of relationships already between schools in the US and those in peace corps countries.
• students are building cross-cultural and spatial reasoning skills by editing map data, which can help peace corps volunteers deliver data to their communities
• will be taking program to a national level over the next few years

Challenges

• building capacity of volunteers and counterparts in a decentalized organization
• mobilizing volunteers with little or no internet access
• motivating key agency stakeholders
• data quality

Up and Coming

• conitinued development and promotion of OSM toolkits
• integrating open map kit from Red Cross
• in-country OSM trainings
• trainings of former volunteers in OSM for students
• OSM certification and badging

Akiko Harayama, OSM at the UN

How is the UN using maps?

• UN Peace Keepers using maps to properly navigate the countries
• Administrators to make high-level decisions
• To inform and advocate, sharing thematic information about countries
• Using OSM for disaster preparedness in Indonesia, you can watch the video about this project on youtube
• Using OSM for proper routing in West Africa, youtube
• global initiative "Humanitarian Data Exchange" that makes humanitarian data easy to find, analyze, and share
• using OSM as a database for their spatial information
• useful during emergency: was used as a dashboard to view statistics during Nepal earthquake
• over 200 datasets to download

BOF: OSM & Government

• USAID using OSM data for crisis mapping and emergency response
• being able to run OSM offline has given it a strong advantage
• NYC community boards to potentially use OSM - looking for good examples of how it is used for local community governments
• France: OSM data are used by police departments, statistical institutes. Local authorities are typically using OSM as basemaps and for some local data. If they don't have data on their open data portal, cities can take data from OSM and use it
• It being "free and open" can be scary, and doesn't necessarily build trust. How can we fix this? Build use-cases and highlight success stories - show how it is better quality data.
• The problem is more "anybody can edit it" - starting to tag edits by specific entities (i.e. DOT data source). A good response is "anybody can fix it!"
• With proprietary data, whether or not if you accept the data as authoritative, you can source it to someone else, rather than a nebulous community

Olaf Veerman, OSM as a Platform

Work with Philippines

• OpenRoads, an OSM platform to manage data on the road network and road investment projects (World Bank & Philippines).
• Using: iD, to-fix, OSM oAuth, Data tools, custom editing API, custom dashboard, custom analytics
• This is using the OSM software, not necessarily the data itself. Sharing data models, though - just not in the same instance.
• Use the iD editor and have custom presets to fit the needs of the government.
• hidden UI elements that aren't necessary for their needs
• One of the custom parts is a whole analytics dashboard, which allows them to view road conditions on a municipal level
• Why using a separate infrastructure?
• data doesn't fit OSM
• user permissions / auth needs
• hosting restrictions
• it's going to require a cultural change to get them out of local management systems

Private OSMs

• Cool vision of OSM being this big core dataset acting as the hub in a hub-and-spoke model for open mapping. I.e. StateOSM, GovOSM, NPSOSM, OpenRoadsOSM
• one challenge would be syncing data back into OSM
• OSM is people, data, and great software