Location-based storytelling with
OSM & LocalWiki

How we adapt LocalWiki and OSM tools for our work

Pomin Wu @pm5

OSM since 2012

g0v.tw, openstreetmap.tw

pm5.github.io/talks

4 parts in this talk

1. About Lowiki project
2. LocalWiki
3. OSM Tasking Manager
4. Future

Lowiki

lowiki.org

A crowdsourcing platform for disaster-related knowledge

Crowdsourcing

• ...to collect local knowledge.
• ...to include governemnt as one of the "contributors".

In other word, we use crowdsourcing process to build collaborations between government and local people.

Information is co-edited by NCDR and the community.

In order to achieve this with a relatively small team, we adapted several open source tools for our purpose.

... and we leverage OSM and its community.

LocalWiki

localwiki.org

A grassroots effort to collect, share and open the world's local knowledge

LocalWiki main features

• Regions
• Pages
• Tags
• Maps
• Templates

I'll talk about how we (Lowiki) uses tags, maps, and templates.

Tag

• We use a 5 tags through out Lowiki:
  • community/apartment complex
  • shelter
  • resource center
  • social facility
  • gas/water station
• Users can use any tag they deem adequate as well.
• This way we have tag views with integrated maps.

Map

1. We enhanced the "import from OSM" feature.
2. We added disaster information from government WMS service.

Template

• We use structral templates (tables) for the 5 default tags.
• The structral information can be exported as CSV, TXT, or KML.
• LocalWiki also has built-in JSON-based RESTful API.
• The data structure (tables) is defined directly in templates, so users can change them as needed.

In other word, text-base descriptions are made structral by providing users with better tools to organize them, much like tag system in OpenStreetMap.

Summary on LocalWiki

• From the domain knowledge of disaster experts, we identify at least 5 core facilities in disaster relief.
• LocalWiki is flexible enough for us to build a crowdsourcing process around these core facilities, using tags, maps, and templates.
• A set of simple template of tables helps us make the information structral.

OSM Tasking Manager

Bootstrap with official data

• Government Open Data such as data.gov.tw has a lot of information.
• Not only do they have polygons, but also attributes such as addresses, phone numbers, operator names, opening hours, etc.
• A lot of that are not on OpenStreetMap yet.

Import

We don't import data frequently because:

• Community discussion and review is a lot of work.
• Import itself is stressful.
• Keeping data up-to-date is hard.

However, there are now 8194 records of defibrillator in the offician data.

Collaborate on data import/synchronization

Doing this with OSM Tasking Manager would be:

• Easier.
• Less stressfull.
• Many of the works can be automated.

"Dataset" in OSMTM

We modified OSMTM so that one can:

• Upload a dataset to a Tasking Manager project.
• Split the dataset by tasks.
• Automatically validate tasks that has no new data.
• Automatically invalidate only the tasks having new data.
• Coordinate import works with Tasking Manager workflow.
• Visualize progress.

Results

• Clearly-defined roles and goals.
• Automated diffing and saves works.
• Visualized progress to encourage collaborations.

Future

Some ideas we are having in mind

Data pipeline from Open Data

Automatically update government Open Data to OSMTM.

Open Data ⇒ OSMTM ⇒ OSM

Data pipeline from LocalWiki

What we have now:

OSM ⇒ LocalWiki import

What we can do:

LocalWiki export ⇒ OSMTM ⇒ OSM

Issues report

• Report OSM issues from LocalWiki.
• watermap.teia.tw

LocalWiki ⇒ OSM Note ⇒ OSM

(Slightly more) structured OSM notes

• With hashtag (#lowiki) and key=value pairs.
• Build issue tracking tools on OSM notes.