WikiProject Belgium/Using AGIV Crab data/The AGIV Crab Import Website
The import webpage can be found at http://crab-import.osm.be. It contains a way to query the addresses in Flanders per postal code and per street.
Next to that, the site also offers the possibility to compare CRAB data with OSM data, to see which housenumbers are still missing, or where there could be mistakes.
In this document, we explain the webpage section per section.
The comparision options allow you to control if and how the data from CRAB and OSM is loaded and compared. When you edit one of these options, you always have to reload the page by clicking the "update" button.
Because you can't just compare all 2 million addresses of Flanders, you have to restrict yourself to a postal code. If you don't know the postal code of the area you want to map, you can check the site of bPost.
This field lets you filter only a limited number of streets. You can enter a complete name (like "Dorpsstraat"), or you can use a wildcard (*) to load multiple streets (f.e. D* to load all streets starting with a D). By limiting the streets, the page will load faster, which may be interesting in big cities.
Load OSM data
When you enable this option, the address data will be downloaded directly from OSM and compared with CRAB data. This option is really necessary when you're working in places where a lot of addresses already exist (to see which are lacking). But you can also choose to omit the comparison, which will make the page load faster.
The data comes directly from Overpass API, and thus it takes some minutes before the data is available after you uploaded it to OSM.
When you enter a distance here, the distance will be used in the comparison. When addresses in OSM and CRAB match, but the distance is bigger than the maximum you gave, then the addresses are seen as non-matching after all.
Sometimes, a CRAB position isn't very precise. So a difference of some 30m isn't too rare.
These options change how the data is exported to JOSM. The options are immediately active (you don't have to press the "update" button to activate them).
Include CRAB info
With this option, extra information from CRAB will be loaded. You can use this option to study a certain point, but this CRAB info shouldn't be uploaded to OSM in any case.
The extra info shown is:
- CRAB:herkomst: This is the origin of the CRAB position. This way, you can see that an address with CRAB:herkomst=afgeleidVanVoordeur will have a better position than one with CRAB:herkomst=afgeleidVanPerceelGrb. The first point will be positioned correctly on the front door, while the other will be on the centre of the parcel (and it can possibly be hard to find the building it belongs too). In total, there are 15 classifications, see the leesmij (readme in Dutch)  for more information.
- CRAB:hrnlabels: When there are multiple numbers overlapping, Agiv groups them under a common housenumber label. F.e. when 24 and 26 overlap, both addresses will be present in the database, but they will have "24-26" as housenumber label.
- CRAB:apptnrs and CRAB:busnrs: are the appartment and bus numbers as noted by CRAB. There's no real difference between the two, but they should never appear both on the same address. You can see both lists here to check which one is correct.
Include municipality and postcode
With this option, the tags "addr:city" (the municipality) and "addr:postcode" (the postal code) will be added to the downloaded data. If the municipality and postal-code boundaries are complete and correct, then these tags aren't necessary. However, the tags can be useful to map border-cases, or to correct municipality and postal-code boundaries when they're wrong or not present.
Include apartment numbers
With this option, the apartment and bus numbers from CRAB will be added in OSM under addr:flats (use the option "include CRAB info) to see the actual source of the info).
Do watch out, the notation of apartment and bus numers isn't standardised at all, and because of that, there are also more mistakes in that data. If you activate this option, you'll also have to check every address to see if the numbers seem logical, and survey it in case of doubt.
Since the apartment numbers aren't hard to add later on (can be automated when there's a good data source), adding it now isn't really necessary.
The data section contains a table with all streets, and info about those streets (total number of houses, number of wrong housenumbers, ...).
The table is also interactive, and can be used to load data in JOSM (with JOSM activated and correctly configured).
- Column headers: When you click on the column headers, you can sort the data of that column. Like this, it's simple to find the street with the most housenumbers, or the one with the most wrong housenumbers.
- Summary: The second row is a summary of all data below it. It sums up the data to track your overall progress.
- Data: The actual data is structured with the following columns:
- Name-column: This column just contains the streetnames as they appear in CRAB. By clicking on a name, the data will be loaded from OSM into the active layer in JOSM.
- Full-column: This column shows how many addresses CRAB has for every street. By clicking on a numer, that CRAB data will be loaded in a new layer in JOSM.
- Missing-colum: This column shows how many addresses are missing in OSM compared to CRAB. In most cases, it's trivial to add these addresses to the right buildings in OSM.
- Missing overlapping-column: This column shows how many addresses are missing in OSM, but where CRAB isn't correct either. The overlapping addresses are most likely addresses without good positional information. To map these addresses, you have to be careful, and a local survey is almost always needed.
- Wrong-column: This column contains the addresses in OSM, but for which no matching address has been found in CRAB. This can be because there's a mistake in OSM, or because CRAB is missing some addresses. To fix this column, you'll also have to do a local survey.
- Completeness-column: This column shows you a completeness percentage. Per wrong or missing address, the percentage is lowered. The number is also used to determine the colour of this table row.
- GPX buttons: Below the table, there are 3 GPX buttons. Via these buttons, you can export all remaining addresses in that column to a GPX file, so you can load it on a smartphone or navigation device. This can help to do the local surveys.
The map is comparable to the table, but offers a different overview.
When the OSM data is loaded, the markers will get a colour-code (from red to green in 10 steps). This way, you can estimate how complete an area is.
When you click on the balloons, you'll see the same info as in the table, and the links work similarly.
When the data is loaded in JOSM, there's a possibility you see a warning as a fixme=* message. Here's an overview of the possible messages and their meaning:
- This street contains abbreviations, please try to expand them: Data in OSM shouldn't contain abbreviations, so you should try to expand it. Abbreviations are recognised by the presence of a dot (.), if there are false-positives, you may just ignore this warning.
- Alphabetic bis-numbers should be capitalised (e.g. 15A instead of 15a): In OSM, we try to use the format specified by bPost, so with capitalised bis-numbers and without a space.
- This housenumber has a numeric bis-number with an underscore. Bis-numbers should be noted as 10/1, 10/2, 10/2, ... or 10 bis, 10 ter, ... Please check locally which format fits best: Just as above, for numeric bis-numbers, we also try to use the bPost format. In this case, CRAB always uses an underscore (_), while bPost advises a slash (/) or bis, ter, ... Deciding which of the two formats to use happens best by a local survey.
- This number is not preset in CRAB. It may be a spelling mistake, a non-existing address or an error in CRAB itself: This is the default message for data from the "wrong" column. It can be used to filter out the data when you merged it by accident with another data set, or when you uploaded the data by accident.
- This number contains multiple housenumber labels. As the housenumber labels is a combination of all housenumbers in that location, this is certainly a mistake in CRAB. Please report it to AGIV: This is something that really shouldn't happen at all in the data, some official seriously messed up the data in this case.
- There are both appartment- and busnumbers on this address. Please check what's visible on the front door as part of the address: When you've chosen to also map the bus and apartment numbers, you can see this message. When CRAB data has both sorts of numbers, it's certainly a mistake, and you should probably take a look locally if any one of the two is right.
Note that the messages are not exclusive. So some address node might have multiple mistakes and show multiple messages in one fixme=* tag.