Why do search results favor China?

OK, I’m sitting in Taiwan, looking at Taiwan on the map, and I search for 文曲路,

and the search results are all China, China, China, China, China, China. Only way down at the bottom is the single Taiwan result that I was looking for. Why, why, why, why?

Not surprising. Try 中山路 (Zhongshan Road, one of the most common road names in Taiwan) right on that Zhongshan Road in Banqiao, New Taipei, Taiwan (where the city hall locates) and the results jump to Hebei, Tianjin, China.

What’s your viewport?

1 Like

says 1133 × 577.

1 Like

What is the extent of your map view? When centered on and zoomed out for whole Taiwan, Top 2 results are in Taiwan, and the 4th one. 文曲路 | OpenStreetMap

  1. Way: ‪Wenqu Road‬ (‪226731430‬) | OpenStreetMap
  2. Way: ‪Wenqu Road‬ (‪392573426‬) | OpenStreetMap
  3. Way: ‪文曲路‬ (‪452082230‬) | OpenStreetMap
  4. Way: ‪Wenqu Road‬ (‪392573423‬) | OpenStreetMap

Looking at the debug view, I don’t know what may cause a failure in “apply viewbox”

The unbounded query seems to have equal importance=0.05430 for all almost results. It may be a problem in presentation, not analysis. Nominatim - Debug

Preliminary Results

[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W234975445, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991169, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W836977775, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991202, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991200, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W836977774, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W452082230, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W226731430, cc=tw, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W392573423, cc=tw, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W392573426, cc=tw, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W197114324, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W560704661, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W475500732, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,construction), rank=26, osm=W875327495, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W191476804, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W1160860419, cc=cn, importance=0.05339)

TOTAL: 16
Results after reranking

[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W234975445, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991169, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W836977775, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991202, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991200, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W836977774, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W452082230, cc=cn, importance=0.05340)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W226731430, cc=tw, importance=0.05340)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W392573423, cc=tw, importance=0.05340)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W392573426, cc=tw, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W197114324, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W560704661, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W475500732, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,construction), rank=26, osm=W875327495, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W191476804, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W1160860419, cc=cn, importance=0.05339)

TOTAL: 16
Final Results

[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W234975445, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991202, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991169, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991200, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W836977775, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W836977774, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W452082230, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W197114324, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W560704661, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W475500732, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,construction), rank=26, osm=W875327495, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W191476804, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W1160860419, cc=cn, importance=0.05339)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W226731430, cc=tw, importance=0.05340)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W392573426, cc=tw, importance=0.05340)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W392573423, cc=tw, importance=0.05340)

TOTAL: 16

PRC results only appear on top at z=7 Nominatim Demo

“Final results” only placed a PRC result at #2 compared to a “reranking”, followed by 3 Taiwan results Nominatim - Debug

Results after reranking

[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W452082230, cc=cn, importance=0.05340)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W392573426, cc=tw, importance=0.05340)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W392573423, cc=tw, importance=0.05340)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W226731430, cc=tw, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W1160860419, cc=cn, importance=0.05339)
[0.933]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W234975445, cc=cn, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991202, cc=cn, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991200, cc=cn, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W836977775, cc=cn, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991169, cc=cn, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W836977774, cc=cn, importance=0.05340)

TOTAL: 11
Final Results

[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W452082230, cc=cn, importance=0.05340)
[0.433]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W1160860419, cc=cn, importance=0.05339)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W226731430, cc=tw, importance=0.05340)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W392573426, cc=tw, importance=0.05340)
[0.833]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W392573423, cc=tw, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,residential), rank=26, osm=W234975445, cc=cn, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991202, cc=cn, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991169, cc=cn, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W147991200, cc=cn, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W836977775, cc=cn, importance=0.05340)
[0.933]
    PLACEX(文曲路, type=(highway,tertiary), rank=26, osm=W836977774, cc=cn, importance=0.05340)

TOTAL: 11
2 Likes

I’m guessing it’s caused by lack of exact match from the section number in name=
Maybe would help with a *_name=中山路 ?

Is your browser language set to English first? Then that is likely the cause.

Streets in Taiwan seem to feature mostly a name:en tag, so that you get the English name back in the search result. In China the streets have only a name, so that the result display remains in Chinese letters. Nominatim does a final reranking to match the search query against the results. You put Chinese writing in, you get preferably Chinese writing out. (Note that Nomiantim doesn’t make a difference between simplified and traditional Chinese, which admittedly doesn’t help in this case.)

That said: Nominatim is not Google and doesn’t know about your complete personal history. If you under-specify your search by just putting in a single very common street name, you cannot expect more than random results.

3 Likes

Additional, the writing method of 中山路/文曲路 is same in both S/T Chinese. We can’t determine the 文曲路 is expected to be in Shanghai or in Taipei according to only a user input string. We can’t know.

Another question: Maybe nominatim can get user bbox center and first consider places in such region?


我知道你很急,但你先别急
I know you’re in a hurry, but don’t be in a hurry

That is done as well when you are searching by using the search box on osm.org directly. There just is no viewbox included when going to a search result via the direct link that Jidanni posted.

1 Like

And the results are in!

No matter what my Accept-Language header says, as long as 文曲路 is more than a couple miles off of my screen, whammo, search sends me on a one-way express ticket to China (and most users would have to paw their way back dragging the mouse, unless they try their browser's BACK button.)

The search button should realize that there is a 99% the user is looking for something in the same country the map is currently showing.

Maybe it thinks Taiwan is China?

Well if there was a "[ ] search internationally?" checkbox we could test that theory.

more detailed said, same region or in bbox element should be prefered, for example you might not want to have your 中山路 bring your window to Taichung while you are zooming in Taipei

So I think this is a technical problem rather than a political question, and hope nomimatim’s search for east asian place name can better.

1 Like

Have you tried checking if the lookup logic supports fuzzy matching or alternate name fields (e.g., alt_name= or official_name=)?

1 Like

Officialy what we call Taiwan in occident still claims to be Republic of China (R.O.C. on your visa/stamp). They administer 2 provinces: Taiwan and Fujian (that are limited to some island, at border of People Republic of China, Fujian province. Search kinmen (金門) that is in the bay of xiamen (厦门) for example.

Sun Zhongshan, was the author of the revolution in Wuhan (mainland), that finished the Qing dynasty and started the Republic of China (that is limited to Tawian and fujian islands now). He was the founder of Chinese nationalist party, Kuomintang, the flag of this party is the flag of Taiwan). So You willl find its name every where in China (Both People republic and (nationalist) Republic). The fight between nationalist and communist parties started with Chang Kai-Check, in Shanghai, with huge massacre of workers during a strike, helped by some local mafias. This one finally flee to Taipei in 1949 at the end of civil war, and was its dictator for several decades, with white terror regime, helped by US. It tried to take back mainland without success. You will not found it’s in mainland street names, but its tomb guards show is a national event everydays.

1 Like

That’s a very common name in whole China, not only Taiwan, Sun Zhongshan 孙中山 it’s the main author of Wuchang uprising, in Wuhan, that make the fall of the last Emperor of China. So there are at least one Zhongshan park in all big cities and lot of Zhongshan road everywhere, in mainland as in Taïwan or Fujian islands. Mainland one are maybe requested more often, maybe larger. The best way is to add the city or the district or provincial limit (like taiwan).

1 Like