There is a Chinese account that has normalized the KFC tag that its changeset boundary is so big recently, which included Taiwan in it. The changeset boundary is so big that makes QA tools like OSMCha not function as expected, probably Overpass API limitation. The reviewer has to manully review the changeset if there is anything wrong. As a Taiwanese mapper I check if he change something inside the Taiwan boundary, and foundout he does what he said. The mapper stated in his changeset comment said he did not touch KFC in the “Taiwan area”, which in my humble option is not quite necessary.

I will recommend that even though the cause is QA tools problems, that can not check edit in a country if it is problem changeset, might be a good idea to avoid such a big changeset boundary. Split the big changeset into a much smaller, much checkable changeset that lets QA tools like OSMCha to function normally. This could avoid potiental disruption. Heavy mappeers could focus on mapping, not changeset comment discussion.

Even though Taiwan is a small place, there are many hard-working mappers that contribute to OpenStreetMap, following the truth to map stuff that exists in front of them, that can be observed by their own eyes. But sometimes we have to deal with people who do not follow the on-the-ground truth, forcing Taiwanese to accept things not observabe on the ground.

… See full entry

漢字版本

最近有一个中國來的口座正規化KFC的標籤，舞的範圍有夠大，編輯變動的範圍猶閣有共臺灣包入去，予一寡QA工具攏無法度正常視覺化顯示變啥物魍，應該是 Overpass API 的限制造成的，愛手工檢查是毋是有真真正正共臺灣範圍的物件創到。伊的編輯留話的所在有特別強調無創「臺灣地區」的 KFC，我是感覺這敢有需要特別強調？

建議是講，雖然 QA 工具出問題，無法度確認是毋是有問題的編輯，雖然是弄一个國家範圍的物件，欲共舞的範圍創較細的，分幾若擺來編輯上傳，較袂引起誤解，嘛較簡單會當審核內容敢有問題。有法度閃避遮的問題，𠢕人嘛會當專心畫𪜶所在的物件，毋免予人感覺有問題愛閣開時間溝通討論。

臺灣雖然是小所在，嘛是有真濟辛辛苦苦拍拚骨力，維護圖資的圖客，足頂真依照事實，猶閣有眼前會當看著的物件來畫地圖。毋過不時有人欲來挑戰加上非事實的物件，強逼臺灣遮的人愛接受非在庄在地草地觀察會著的物仔。這个代誌是無遵守 OpenStreetMap 的現地觀察，才會當加入 OpenStreetMap 的 on the ground truth 原則。

OpenStreetMap 親像𪜶的姊妹仔計畫 Wikipedia 仝款，是人人會當看著怪怪的物件，著會使修正，甚至實際上無存在的物仔嘛會當刣掉，毋免有較懸權限的口座來做。甚至較需要了解佮操作智識的 reverter 嘛毋免一定是管理員才會當做，有社群共識一般的口座就會使做。OpenStreetMap 計畫內是較平等的，無講一定欲予較資深的人較有資格，有理就好。

Tâi-lô pán-pún

Tsuè-kūn ū tsi̍t-ê Tiong-kok lâi-ê kháu-tsō chèng-kui-hòa KFC ê piau-tshiam, bú-ê huān-uî ū-kàu-tuā, pian-tsi̍p piàn-tōng ê huān-uî ū kā Tâi-uân pau-ji̍p-khì, hōo tsi̍t-kuá QA kang-kū lóng bô-huat-tōo tsìng-siông sī-kak-huà hián-sī pìnn-siánn-mn̍gh-báng, ing-kai-sī Overpass API ê hān-tsè tsō-sîng-ê, ài tshiú-kang kiám-tsa sī-m̄-sī ū tsin-tsin-tsiànn-tsiànn kā Tâi-uân huān-uî ê mn̍gh-kiānn tshòng-kàu. I-ê pian-tsi̍p lâu-uē ê sóo-tsāi ū ti̍k-pia̍t kiông-tiāu bô-tshòng “Tâi-uân tuē-khu” ê KFC, guá sī kám-kak tsē kám-ū su-iàu ti̍k-pia̍t kiông-tiāu?

Kiàn-gī sī-kóng, sui-jiân QA kang-kū tshut-būn-tê, bô-huat-tōo khak-jīn sī-m̄-sī ū-būn-tê ê pian-tsi̍p, sui-jiân sī lōng tsi̍t-ê kok-ka huān-uî ê mn̍gh-kiānn, bueh kā bú-ê huān-uî tshòng khah-suè–ê, hun kuí-nā-pái lâi pian-tsi̍p siōng-thuân, khah-buē ín-khí gōo-kái, mā khah-kán-tan ē-tàng sím-hi̍k luē-iông kám-ū būn-tê. Ū-huat-tōo siám-phiah tsia-ê būn-tê, gâu-lâng mā ē-tàng tsuan-sim uē in sóo-tsāi ê mn̍gh-kiānn, m̄-bián hőng kám-kak ū-būn-tê ài koh khui-sî-kan koo-thong thó-lūn.

… See full entry

中國那邊有人在日記抱怨有大量學生畫中南大學一帶，看起來是老師要求學生畫學校附近，但由於缺少 QA 程序，因此產生不少問題資料無人處理。這種事情早晚都需要面對，需要動員社群成員去修正，或是決定資料實在是太差了，通通砍掉處理。極端嚴重的情形可以找 Data Working Group 協助，將溝通無效的帳號封一段時間，給予社群時間來修正。

臺灣這邊的經驗則告訴我們，要與顯然是做老師指定的作業學生溝通，用變更集留言，大都是沒什麼回應的。社群有時間就直接介入修正就好，再來研究他們學生是從那裡來的。如果沒時間也不想幫人修正，那就通通刪掉就好，必要時再請 Data Working Group 處理，能夠暫時封掉沒有回應的帳號，暫緩不讓問題擴大。

我還記得臺灣這邊遇過最嚴重的事情是美光團，他們用 HOT Task Manager 畫彰化、雲林一帶的建築，最後實在產生太多問題建築，已經超出社群能修正的程度，與 Data Working Group 談好就把他們產生的資料通通刪除。

其他遇到的問題編輯團體還有彰化學校的編輯團，一直到現在都還不知道是那間學校的學生所為。還有一個加心理醫療場所的人，也是有問題，因為八成他的經緯度是從 Google 地圖用 Geo-coding 產生的，資料授權是有問題的。

OSM Wiki 有組織編輯指引，當有團體計劃有大量成員編輯時，就要在 Wiki 頁面登記，要留帶頭的聯絡人資料，畫地圖的目的，以及持續的時間。臺灣環境資訊協會的飲水地圖計畫就有留資料登記。組織編輯指引是 OpenStreetMap 生態系相當重要的指引，也能看其他團體的範例，是怎麼登記他們計劃做的事情。

對於社群來說，找到帶頭者，跟老師能夠好好溝通，老師得知道 OpenStreetMap 是不少重要網站依賴的圖資，學生投入貢獻需要認真，不能隨意增添地圖資料。對於剛接觸 OpenStreetMap 的學生來說，先讓他們畫簡單的東西，例如學校、住家附近的店家，別一下子給學生太過複雜，範圍太大的事務。

再來中國的環境其實不允許一般人來畫地圖，個人畫地圖是違反中國的法律。找到地圖團的帶頭者，然後在組織編輯頁面登記，好像那裡怪怪的，並不是好的處理方式。另外學生帳號用真實姓名的問題，在中國的網路環境下，還是用代號，別輕易曝光真實姓名比較好。

China mappers complain about there are many students mapping around Zhongnan University, which seems assigned by school teachers, but lack of QA process after students had done their mapping tasks. These kinds of issues are faced by the community someday, whatever the community decides to fix it, or just remove the low-quality stuff, it is all OK. And in an extreme case, the community could ask the Data Working Group for help, by banning accounts that keep editing and adding low-quality data, getting some precious time for the community members to fix the data.

As a Taiwanese mapper, Taiwan experience of dealing with school assigned tasks from teachers, leaving changeset comments are quite useless, few will response to the comments. Directly fixing the error or low-quality data is faster, and reseach where do these change come from? If you do not want to spend your time of fixing error or improving the data quality, then just delete the stuff is OK. If the damage so big and no response from the mapper, can’t be handle by community, then the Data Working Group can step in and no further damage.

Recenlty most serve incident in Taiwan is Micron editing group. They use HOT Task Manager to draw buildings in Changhua and Yunlin. They drew so many problem buildings, that can’t be handle by the community, so the HOT ask Data Working Group to step in and delete all the buildings by Micron editng group.

Another incidents like editing group from shcool of Changhua, which one could not trace which school they belong to. And another incident like some mappers adding psychological medical clinic, which using coordinate geocoding by Google Maps, there is a data license issue.

… See full entry

漢字版

中國遐有人寫日記哼(hainn)有大量學生佇中南大學畫地圖，看起來是老師按學生畫學校附近，但是缺少 QA 的程序，予袂少有問題的資料無人插。這是早慢欲面對的問題，社群愛動員去修理，抑是畫甲傷䆀做理由攏刣刣掉矣。嘛會使揣 Data Working Group 協助，共講袂聽的口座封一寡時間，予社群成員有時間清理佮修正。

臺灣遮的經驗是，共一寡看起來是學生做學校指定作業的口座溝通，用變更集留話，通常是無回應的。對社群來講有時間就直接修改好勢，閣來研究𪜶是對佗位來的。無想欲改就攏總刣掉嘛無要緊，必要時予 Data Working Group 處理，處理無回應閣繼續編織的口座，至少無予問題擴大。

會記得臺灣遮上嚴重的底代是美光團，用 HOT Task Manager 佇咧畫彰化，雲林遐的建築，最後因為問題建築傷濟矣，無法度修改好勢，予 Data Working Group 資料攏刣刣掉。

其他的代猶閣有彰化遐學校學生的編輯團，一直毋知是佗位的學校咧畫。猶有人加上心理醫療場所的POI，嘛是有一寡問題，因為應該是共地址用 Google 轉成經維度，資料的授權有淡薄仔問題。

OSM Wiki有組織編輯指引，有團體愛編輯，著愛登記佇 Wiki 頁面，愛留𤆬頭的人聯絡資料，目的，佮時間。臺灣環境資訊協會的啉水地圖計畫就有留資料登記。組織編輯指引應該是 OpenStreetMap 生態系內底，會當參考佮遵守的指引，嘛會當參考別人是按怎處理的。

上好是揣著頭，共老師好好矣溝通，老師愛知影 OpenStreetMap 予袂少重要的網站使用，學生貢獻是愛頂真，袂使清彩處理。通常予學生畫較簡單的事物，親像講學校，徛家附近的店家，莫予學生畫傷複雜，傷大的範圍。

又閣有中國的環境無允准一般人來畫地圖，個人畫地圖是違反𪜶的法律。愛揣到頭人佇Orgaizated Editing Guideline頁面登記，感覺毋是傷好的處理方式。另外學生個人號口座用真實姓名抑是問題。佇中國的網路環境，線頂的口座號一个代號較好，莫用真實姓名號口座的名號是較好的做法。

Tâi-lô-pán

Tiong-kok hiah ū-lâng siā ji̍t-kì hainn ū tuā-liōng ha̍k-sing tī Tiong-lâm Tāi-ha̍k uē tuē-tôo, khuànn-khí-lâi sī lāu-su àn ha̍k-sing uē ha̍k-hāu hù-kūn. Tse-sī siā-kûn tsá-bān bueh bīn-tuì ê būn-tê, siā-kûn ài tóng-guân khì siu-lí, ia̍h-sī uē kah siunn-bái tsuè-lí-iû lóng thài-thài-tiāu ah. Mā ē-sái tshē Data Working Group hia̍p-tsōo, kā kóng-buē-thiann ê kháu-tsō hong tsi̍t-kuá sî-kan, hōo siā-kûn sîng-guân ū sî-kan tshing-lí kah siu-tsìng.

Tâi-uân tsia ê king-giām sī, kā tsi̍t-kuá khuànn-khí-lâi sī ha̍k-sing tsuè ha̍k-hāu tsí-tīng tsok-gia̍p ê kháu-tsō kau-thong, iōng piàn-king-tsi̍p liû-uē, thong-siông sī bô huê-ìng ê. Ū-sî-kan tō ti̍t-tsiap siu-kái hó-sè, koh-lâi gián-kiù in tuì tó-uī lâi ê. Bô siūnn bueh kái tō lóng thâi-tiāu mā bô iàu-kín, pit-iàu sī hōo Data Working Group tshú-lí, tshú-lí bô huê-ìng koh kè-sio̍k pian-tsi̍p ê kháu-tsō, tsì-tsió bô hōo būn-tê khok-tāi.

… See full entry

漢字的版本

最近毋知是按怎，近期的 zh-CN 日記出現誠濟篇指控 OpenStreetMap 無遵守中國的法律，竟然共台灣標示是獨立的國家。

臺灣人看著這款指控真無爽快，不而過猶是有精神來回應，是講 OpenStreetMap 的政策、運作方式佮精神，特別是爭議領土欲按怎處理的方式。

Pe̍h-ōe-jī ê pán-pún

Chòe-kūn m̄-chai sī án-chóaⁿ, kūn-kî ê zh-CN ê ji̍t-kì chhut-hiān chiâⁿ-chōe phiⁿ chí-khòng OpenStreetMap bô chun-siú Tiong-kok ê hoat-lu̍t, kèng-jiân kā Tâi-oân piau-sī sī to̍k-li̍p ê kok-ka.

Tâi-oân-lâng khòaⁿ-tio̍h chit-khoán ê chí-khòng chin-bô sóng-khoài, put-lî-kò iáu-sī ū-cheng-sîn lâi hê-èng, sī-kóng OpenStreetMap ê chèng-chhek, ūn-chok hong-sek kah cheng-sîn, tek-pia̍t sī cheng-gī léng-thó͘ boeh án-chóaⁿ chhú-lí ê hong-sek.

Tâi-lô ê pán-pún

Tsuè-kūn m̄-tsai sī án-tsuánn, kûn-ki ê zh-CN ê ji̍t-kì tshut-hiān tsiânn-tsuē phinn tsí-khòng OpenStreetMap bô tsun-siú Tiong-kok ê huat-lu̍t, kìng-jiân kā Tâi-uân piáu-sī sī to̍k-li̍p ê kok-ka.

Tâi-uân-lâng khuànn-tio̍h tsit-khuán ê tsí-khòng tsin-bô sóng-khuài, put-jî-kò iáu-sī ū tsing-sîn lâi hê-ìng, sī-kóng OpenStreetMap ê tsìng-tshik, ūn-tsok hong-sik kah tsing-sîn, ti̍k-pia̍t sī tsing-gī líng-thóo bueh án-tsuánn tshú-lí ê hong-sik.

Pe̍h-ōe-jī Pán-pún

Tâi-oân chhun-lí ê chu-liāu in-ūi Wikidata siā-kûn chhú-lí hōe-ji̍p ê sî-chūn, the̍h Chú-kè-chhù 2017-nî ê chu-liāu lâi hōe-ji̍p, chō-sêng 2018-nî sin-tiâu-chèng ê chhun-lí, chhin-chhiūⁿ sī Tâi-lâm-chhī hêng-chèng-khu tōa-chéng-pèng, iáu-koh-ū in-ūi jîn-kháu chōe só͘-í hun-thiah ê lí, pēng bô sin-cheng kàu Wikidata téng-koân. Che tō chō-sêng tī OpenStreetMap chit-pêng ū chhun-lí koan-hē tùi-èng bōe-tio̍h Wikidata ê chōng-hóng.

kóng-tio̍h OpenStreetMap chit-pêng, tong-chho͘ Chìn-hoân sī ēng Lōe-chèng-pō͘ Kok-thó͘ Chhek-hōe Tiong-sim sek-hòng ê shape tóng-àn, lâi chhú-lí tùi-èng Hō͘-e̍k-chèng Hē-thóng Tāi-bé, iáu-koh-ū āu-sio̍k tùi-èng Wikidata sî, chhái-iōng pí-kàu sin jî-chhiáⁿ keng-sin khah kín ê chu-liāu, in-chhú tī OpenStreetMap chia kan-ta chha 一个 2021-nî 2-ge̍h chhoe-it sêng-li̍p ê lí, sī tī-teh Hûn-lîm-koān Táu-la̍k-chhī ê Chèng-sim-lí, só͘-í būn-tê bô-tōa, sin-cheng-ka koan-hē tō ē-sái lah.

Hôe kàu Wikidata chit-pêng, 2019-nî khai-sí ū-chi̍t-kóa Wikidata ê chham-ú-chiá khai-sí hōe-ji̍p chhun-lí, siông-sòe khòaⁿ chhun-lí hāng-bo̍k ka ê chham-khó-chu-liāu, sī ēng Chú-kè-chhù ê chhun-lí tùi-èng Hō͘-e̍k-chèng hē-thóng tāi-bé ê tóng-àn. Chú-kè-chhù ê tóng-àn kan-ta keng-sin kàu 2017-nî 1-ge̍h 31-hō, í-āu tō bô keng-sin lah, ti̍t-chiap iōng Lōe-chèng-pō͘ ê Hō͘-e̍k-chèng hē-thóng tāi-bé. Chin-chōe-lâng chōe chhun-lí hōe-ji̍p ê sî-chūn iōng-tio̍h kū ê chu-liāu, chō-sêng āu-lâi sin-cheng-ka ê chhun-lí pēng bô hōe-ji̍p kàu Wikidata téng-koân, kè-kiong ū 75-ê sin-cheng-ka ê chhun-lí, pēng-bô siu-lio̍k kàu Wikidata.

Wikidata chit-pêng chòe hōe-ji̍p ê lâng, 2020-nî lōng chhun-lí, mā bô-chù-ì tio̍h hit-sî-chūn ū chi̍t-kóa chhun-lí í-keng chéng-pèng siau-sit leh, m̄-koh in-ūi chhái-iōng chú-kè-chhù kū-ê chu-liāu só͘-í bô-hoat-tō͘ hoat-kak kàu būn-tê. Lán mā ē-tàng khòaⁿ-tio̍h chia-ê hōe-ji̍p chhun-lí hāng-bo̍k khiàm-chió Eng-bûn ê chōng-hóng.

… See full entry

PE̍H-ŌE-JĪ pán-pún:

Góa tī-teh MapComplete Telegram kûn-cho͘ ê liû-giân:

Kí-gōa nî-chêng, Tâi-oân ū-chi̍t-kóa tô͘-kheh khai-sí Lim-chúi tōe-tô͘, chia-ê tô͘-kheh put-chí-sī piáu-sī lim-chúi-tiám, mā siūⁿ-boeh kā Lim-chúi tōe-tô͘ bāng-chām ê têng-sek-bé, ōaⁿ pa̍t-ê chú-tê thui-chhut lēng-gōa ê chú-tê tōe-tô͘, lâi thui-sak Khai-hóng kòe-tō͘. M̄-koh chiâⁿ khó-sioh bô-hoat-tō͘ the̍h-tio̍h chu-goân lâi thui-sak bô-kâng ê chú-tê tōe-tô͘, khip-ín Tâi-oân-lâng lâi chham-ú Khai-hóng koe-tô͘.

Chin-hoaⁿ-hí MapComplete ē-tàng tit-tio̍h Khai-hóng Koe-tô͘ Ki-kim-hōe, Khai-hóng Tì-sek Pí-lī-sî, Trage Wegen, OSGeo Belgium, Jîn-tō Kiù-oān Khai-hóng Koe-tô͘ Sió-cho͘, Ûi-sit Tōe-tô͘ téng-téng ê chi-oān, ē-tàng hō͘ kong-iōng pán-pún ê chú-tê tōe-tô͘ 成為 chi̍t-chióng khó-lêng.

Chìn-chêng choān-siá Lim-chúi Tōe-tô͘ siong-koan ê bûn-chiuⁿ, ē-tàng khah chheng-chhó liáu-kái siā-kûn kah NGO chih-chiap ê būn-tê: 1, 2

Tâi-lô pán-pún:

Guá tī-teh MapComplete Telegram kûn-tsoo ê liû-giân:

Kí-guā nî-tsîng, Tâi-uân ū-tsi̍t-kuá tôo-kheh khai-sí Lim-tsuí tuē-tôo, tsia-ê tôo-kheh put-tsí-sī piáu-sī lim-tsuí-tiám, mā siūnn-bueh kā Lim-tsuí tuē-tôo bāng-tsām ê tîng-sik-bé, uānn pa̍t-ê tsú-tê thui-tshut līng-guā ê tsú-tê tuē-tôo, lâi thui-sak Khai-hóng kuè-tōo. M̄-koh tsiânn khó-sioh bô-huat-tōo the̍h-tio̍h tsu-guân lâi thui-sak bô-kâng ê tsú-tê tuē-tôo, khip-ín Tâi-uân-lâng lâi tsham-ú Khai-hóng kue-tôo.

Tsin-huann-hí MapComplete ē-tàng tit-tio̍h Khai-hóng Kue-tôo Ki-kim-huē, Khai-hóng Tì-sik Pí-lī-sî, Trage Wegen, OSGeo Belgium, Jîn-tō Kiù-uān Khai-hóng Kue-tôo Sió-tsoo, Uî-sit Tuē-tôo tíng-tíng ê tsi-uān, ē-tàng hōo kong-iōng pán-pún ê tsú-tê tuē-tôo 成為 tsi̍t-tsióng khó-lîng.

Tsìn-tsîng tsuān-siá Lim-tsuí Tuē-tôo siong-kuan ê bûn-tsiunn, ē-tàng khah tshing-tshó liáu-kái siā-kûn kah NGO tsih-tsiap ê būn-tê: 1, 2

漢字版本：

我佇咧 MapComplete 群組的留言：

幾偌年前，台灣有一寡圖客開始啉水地圖，遮的圖客希望不止是標示啉水點，嘛想共佮啉水地圖網站的程式碼，換別的主題推出另外的主題地圖，來推捒開放街圖。毋過誠可惜無法度提著資源來推捒無仝的主題地圖，吸引台灣人來參與開放街圖。

… See full entry

我在 MapComplete Telegram Group 留言：

很多年前，台灣有些圖客開始飲水地圖計劃，這些圖客不只想要標示飲地點，還要把製作飲水地圖網站的程式碼，換不同的主題來推出另一個主題地圖，藉此推廣開放街圖。但很可惜無法拿到足夠資源來推展不同主地圖，來吸引台灣大眾參與開放街圖。

很高興 MapComplete 得到開放街圖基金會、開放知識比利時、Trage Wegen、OSGeo Belgium、人道救援開放街圖小組、遺失地圖等等的的支援，讓公用版本的主題地圖成為可能。

先前撰寫有關飲水地圖的文章，可以較較清楚瞭解社群與 NGO 接觸遇到的問題：1、2

Due to ill-considered from the Wikidata community using old and outdate 2017 dataset from Directorate General of Budget, Accounting and Statistics (DGBAS), Executive Yuan, Taiwan, the newly found villages in 2018, like the villages mass modification in Taiwan in 2018, or split villages due to increase population, these new villages was not included and imported into Wikidata. So there are some OpenStreetMap village relations could not link to Wikidata.

For the OpenStreetMap part, typebook using relative new and up-to-date shape files from National Land Survaying and Mapping Center, Ministry of Interior Affairs, to add the Code of Household Registration and Conscription Information System (HRCIS Code) and the followup Wikidata item. There are only mssing Zhengxin Village, Dauliu City, Yulin County on OpenStreetMap. OpenStreetMap Taiwan community could just add relation to fix the problem on OpenStreetMap.

For the Wikidata part, there are several users start to import villages data to Wikidata, which started from 2019. If you look closer to some of the Wikidata items, the reference they used link to files from the DGBAS, which is out-of-data and the last update time is 2017-01/31. DGBAS decide to switch to use the HRCIS Code. It indicated that Wikidata users using old dataset to import villages, so the 75 new villages was not included, is missing on Wikidata database.

When Wikidata Taiwan community start to deal with the villages data in 2020, there are many villages already combined in disappear. But due to the fact they used DGBAS out-of-date dataset, they could not found out their are missing mass amount of villages. The DGBAS also missing the villages English name, so we see the imported item missing English label.

… See full entry

Some words that I left at the MapComplete Telegram Group:

Several years ago, there are some mappers start a drinking water map in Taiwan. These mappers are planing to change the map subject, using the code of the drinking water map to deploy more theme map to promote OpenStreetMap. But unfortunately we couldn’t get the resource to make it possible to get plenty of theme maps to get Taiwanese people interested.

Glad to see you guys got supported from OpenStreetMap Foundation, Open Knowledge Belgium, Trage Wegen, OSGeo Belgium, Humanitarian OpenStreetMap Team, Missing Maps.

Some background information about the Drinking Water Map: 1, 2.

最近不知道怎麼了，近期 zh-CN 日記出現很多篇指控 OpenStreetMap 沒有遵守中國法律，標示台灣為獨立國家。

對於台灣人來說，看到這些指控很不舒服，不過還是打起精神來回應指出 OpenStreetMap 的政策、運作方式與精神，特別是爭議領土的處理方式。

台灣村里資料由於Wikidata社群處理上傳匯入時，由於拿主計處2017年資料來匯入，導致2018年新調整的村里，像是台南市里行政區大整併，還有一些因為人口多而分拆的里，並沒有新增到Wikidata。這導致在OpenStreetMap這邊有村里關係對應不到Wikidata。

至於在OpenStreetMap這邊，由於當初晉帆是用內政部國土測繪中心釋出的shape檔案，來處理對應戶役政系統代碼，還有後續對應Wikidata，採用比較新而且更新快的資料，因此在OpenStreetMap只差了一個2021 02/01成立的里，為雲林縣斗六市正心里，問題因此不大，新增關係就可以。

回到Wikidata這邊，2019年開始有Wikidata參與者匯入村里，細看一些村里項目用的參考文章，是用主計處的村里對應戶役政系統代碼的檔案。主計處檔案只更新到2017-01/31，之後就沒在更新，直接用內政部的戶役政系統代表。許多做村里匯入的人用到舊資料，導致後來新增的村里並不會匯入到Wikidata上面，總計有75個之後新增的村里，並沒有收錄到Wikidata。

Wikidata這邊做匯入的人，2020年弄村里，卻有不少村里那時候早已經整併消失了，卻因為採用主計處檔案而沒有察覺到問題。由於主計處檔案沒有附上英文名稱，我們也能在這些匯入的村里項目看到缺少英文標籤的狀況。

因此有兩件事情需要做，第一件事情是將Wikidata上缺漏的另一個是將整併的村里，在Wikidata標示整併的日期。在第一件事情完成後，OpenStreetMap社群才能為這些沒有Wikidata連結的村里關係加Wikidata連結。

未來還有屏東市與東港鎮的新增里，看起來會用新的戶役政系統代碼，OpenStreetMap由於是反映即時狀況，需要等到成立的那天才能編輯更新，不過Wikidata卻不是，可以預先編輯預定新增的里。目前先用hackmd撰寫台灣村里整理 (2021)，協同OpenStreetMap與Wikidata台灣社群可以做的事情。

During the OpenStreetMap x Wikidata monthly meetup, we decide the person who abuse the data from Facebook MapwithAI project, the user whose account starts with c, his contribution of road data that has not been touched will all be removed. MapwithAI have trouble recognized dry creek as road, or recognized high-voltage power line as road. Manual judge by human mappers is still needed.

Despite complains from OpenStreetMap mappers, there are Rudy Map users or foreign hikers in Taiwan complain about strange mountain road that seems appears from no where. It is time that OpenStreetMap communities should interfere, otherwise that will be huge amount of low quality data import into OpenStreetMap.

We have to observate that person if he will be upset and revert back his edit. If this is the case, it is time that Data Working Group to deal with.

週一的月聚上，決定把濫用 MapwithAI 資料的 c 開頭使用者，他編輯的資料只要沒有動過的，版本是 1 的道路都移除掉。太多把乾溪谷辨識為道路的狀況。而在都市裡頭，則是沒有建築，陰影處辨識為巷弄，但實際上並沒有巷弄，比對 NLSC Open Data 版本之後也沒發現有巷弄。

繼續查下去，上次他在 12/31 的編輯也弄了一堆 MapwithAI 的道路資料進來，也該好好檢視狀況。他怎麼那麼有自信 AI 判斷都對，已經有不少狀況如前面提及的乾溪谷，還有高壓電塔下面，都會誤判是道路。需要人工判斷是否是道路，也是人類編輯需要介入的時機。

目前除了圖客的抱怨之外，已經有魯地圖的使用者，以及在台灣爬山的外國人，抱怨 OpenStreetMap 圖資出現奇怪無來由的山區道路。這時候 OpenStreetMap 社群需要介入，免得誤判的資料大量進來 OpenStreetMap，社群有限的人力無法好好審核導致圖資品質大幅下降。

接著就看這位人兄，是否會不爽然後又恢復他的編輯，如果到那一步就有 Data Working Group 介入的時機了。

The one who solved the mystery of Taiwan Power Company grid, Dan Jacobson noticed that the dataset of Taiwan Power Company’s power pole is added with an additional field of coordinates. He announced his findings in OpenStreetMap Taiwan Facebook Group and Talk-TW mailing list.

But the data is released with TWD67 coordinates with special parameters, could not transform directly into WGS84. Thanks typebook for the help to transform the dataset into WGS84. It will make further works much easier.

Jacobson asked Taiwan Power company about the coordinates. He said that Taiwan Power Company will add the dataset with TWD96 coordinates. But we don’t know the exactly update date.

There are many dataset with coordinates released on Taiwan governments’ open data portals. These dataset includes police station, freeway ETC toll gates location, YouBike station is imported into OpenStreetMap. But for foundational dataset like address, there are only Tainan and Taichung address available. We Taiwanese Mappers want the whole country’s address dataset is released in the future.

先前破解台電電力座標之迷的積丹尼，注意到台電的電桿資料加上座標了，並且在 Facebook 社團與 Talk-TW 郵件論壇談這件情。不過台電他們釋出的資料採用的 TWD67 套用特別的參數，導致無法直接轉為 WGS84。還好有 typebook 協助轉成 WGS84 格式，要進行後續匯入方便許多。

積丹尼也詢問台電那邊，會在之後資料集釋出時，附上 TWD97 的座標，只是就不知道何時會進行。

台灣的開放資料當中，有不少有座標的資料集，先前就有派出所、高速公路 ETC 收費桿、YouBike 站匯入 OpenStreetMap。不過比較偏基礎圖資的地址，還只有台南市與台中市釋出來，希望有一天全台灣的地址能夠釋出來。

中國人在 OSM 日記抱怨 OpenStreetMap 沒有遵循中國法規，沒有「正確」標示藏南、台灣是中國的，沒有遵循中國的一中政策。看完全沒有編輯記錄也許是假帳號吧，然後引來 Data Working Group，一位中國圖客，以及我一位台灣人解釋實際運作方式，覺得蠻好笑的。

我知道 [Mapbox CN] 有「因地制宜」弄符合中國法國版本地圖，那時看到 weeklyOSM 有報導，貼到台灣的 Facebook 社團 OpenStreetMap 台灣，引發眾人不滿。Mapbox CN 的產業也許是這位人兄該去用的。

身為台灣人，常常要應對這類中國人的質疑，有時他們會真的動手修改不符合他們意識型態的東西，修改回來是不難，三不五時要修改回來有時也不勝其擾的。

從這一篇日記文來看，GNS 資料就有錯誤的情形。

One of my friend says the Taiwanese technology review website Daodu have a article and Podcast discuss about OpenStreetMap, which is actually the discusion brought by Joe Morrison, reviewing big companies like Apple, Facebook, Microsoft have an unofficial alley fighting against Google. For the technology issue, OpenStreetMap is quite a small minority topic. I think if you are not in the geoinformation industrial, and does not read mass amount of technology news, you will not be familiar with OpenStreetMap. Anyway it is good thing to have some coverage in Taiwan.

In my personal option the Daodu article is quite shallow, especially the most important part which give OpenStreetMap its unique standing point, is that united big technology companies like Microsoft, Apple, Facebook, all decide to support OpenStreetMap project by different methods to avoid using Google Maps. If you do not have your own map solution, you have to use the market’s best solution, that is your enemy’s product Google Maps. So these companies accumulate data on OpenStreetMap, supporting the open source map solution, and they do not need to make a map solution on scratch.

Due to Taiwan media environment, media outlet have editors monitor English media to ediit and translate news reports into Mandarin news articles. Daodu used materials that was already discuss for a while in OpenStreetMap ecosystem, and did not follow the activities of local communities in Taiwan. The community in Taiwan is more emphasizing on basic infrastructure, their work is not that easy to get attention on the internet.

For those who want know more about OpenStreetMap project in Taiwan, you could search for the keyword, there are some articles on the first page of search result. I once worked for a technology media, sometimes have the chance to write about OpenStreetMap and the community.

… See full entry