Tsallake zuwa babban abun ciki
Fassarar na'ura
Wani ɗan asalin yaren bai duba wannan shafin ba tukunna. Tsarin kalmomi na iya zama ba daidai ba kuma kalmomin fasaha na iya zama kuskure. Ana maraba da gyare-gyare ta hanyar menu na Ba da gudummawa.

Samfurin fom ɗin amincewa

Bita ta ƙarshe: 2026-07-07.

Sanarwa

Wannan wurin farawa ne don gina tsarin amincewa na musamman ga aiki, ba fom ɗin da lauya ya duba wanda ke shirye don amfani kai tsaye ba. Bukatun kariyar bayanai (data protection) da amincewa sun bambanta dangane da yankin shari'a (jurisdiction). Kowane aiki dole ne ya daidaita wannan samfurin zuwa ƙasar da dokar kariyar bayananta ke aiki, zuwa takamaiman nau'in bayanan da ake tattarawa, da kuma zuwa al'ummar da ake neman amincewarsu, tare da shawarar doka daga ƙwararru a yankin shari'ar da abin ya shafa. Duba babin legal, consent, and community IP don tsarin da wannan samfurin ke aiwatarwa.

Fom ɗin a rubuce + tsarin magana da baki + ƙarin amincewar al'umma, sune abubuwan da littafin jagorar ke ɗauka a matsayin mafi ƙarancin abin da zai yi aiki don tattara bayanan Sarrafa Harshen Jama'a (NLP) na harsunan Afirka tare da masu bayar da gudummawa waɗanda ba su iya karatu da rubutu ba, ayyukan ɗaukar muryoyi, da rumbun bayanai (corpora) da aka samo daga al'umma. Kwafi shi, fassara shi, daidaita shi, kuma yi amfani da shi a matsayin wani ɓangare na tsarin amincewa da aka yi wa bitar ɗa'a.

Yadda za a yi amfani da wannan samfurin​

Amincewa tsarin aiki (workflow) ne, ba fom ba. Tsarin aikin yana da sassa uku, waɗanda dole ne a tsara su gaba ɗaya:

  1. Fom ɗin a rubuce: don masu bayar da gudummawa masu ilimin boko, don adana bayanai, da kuma don hanyar bincike (audit trail) na aikin kansa. Sashe na A a ƙasa.
  2. Tsarin magana da baki: don masu bayar da gudummawa waɗanda ba su iya karatu da rubutu ba, don mambobin al'umma waɗanda suka fi son amincewa da baki ba tare da la'akari da ilimin boko ba, da kuma yanayin da fom ɗin a rubuce bai dace ba. Sashe na B a ƙasa.
  3. Ƙarin amincewar al'umma: don yarjejeniyar haɗin gwiwa ta al'ummar da harshensu, muryoyinsu, ko iliminsu shine abin da ake tattarawa. Sashe na C a ƙasa.

Kowane aiki yana buƙatar dukkan sassan uku. Tsallake kowane ɗayansu shine hanyar gazawar da wannan samfurin ya wanzu don hanawa.

Yi amfani da tsarin aikin kafin ka tattara kowane bayani daga mutane, ko rakodi ne, hotuna, fassarori, ko lakabi (annotations). Amincewar da aka tattara bayan an riga an karɓi bayanai ba kasafai take zama mai inganci ba.

Kafin ka kwafi wannan samfurin​

  • Nemi lauya ƙwararre a yankin shari'ar da abin ya shafa ya duba fasalin da aka daidaita. Dokokin kariyar bayanan Afirka sun bambanta sosai kan yadda ingantacciyar amincewa take (duba country reference); fom ɗin da ya wuce bita a Najeriya yana iya ƙin wucewa a Afirka ta Kudu.
  • Fassara kowane sashe zuwa ainihin harshen mai bayar da gudummawa. Fasalolin Ingilishi ko Faransanci don bayanan aikin ne kawai, ba don fahimtar mai bayar da gudummawa ba. Amincewa da baki a cikin yaren haɗin gwiwa (lingua franca) alhali mai bayar da gudummawa ya fi gane yaren gida ba amincewa ba ce mai cikakken bayani.
  • Saka mai bita da al'umma ta zaɓa a matakin tsari, ba kawai a matakin bincike ba. Abin da ake fahimta a matsayin "amincewa" ya bambanta a tsakanin al'ummomi; abin da ke zama kyakkyawan tsarin haƙƙin mallakar fasaha (IP) ya bambanta a tsakanin al'ummomi; duka biyun suna buƙatar gudummawar al'umma a lokacin tsara aikin.
  • Adana bayanan amincewar da aka kammala tare da bayanan da aka tattara. Rakodin amincewa na sauti, fom ɗin da aka sanya hannu, bayanan taron amincewar al'umma: duk suna tafiya tare da rumbun bayanan (corpus) domin masu bita na gaba su iya tantancewa.

Kwafi sashen da ke ƙasa zuwa sabon daftarin aiki, fassara shi zuwa ainihin harshen mai bayar da gudummawa, maye gurbin kowane [BRACKETED FIELD] da ainihin amsar, kuma yi amfani da shi don masu bayar da gudummawa masu ilimin boko da kuma don bayanan aikin.

Fasali: [X.Y] · Kwanan wata: [YYYY-MM-DD] · Harshe(n) wannan fom: [LANGUAGES]

Game da wannan aikin​

[PROJECT NAME] aiki ne da [PROJECT ORGANISATION] ke gudanarwa don [ONE-SENTENCE PLAIN-LANGUAGE PURPOSE, NOT "TO ADVANCE NLP" BUT SOMETHING LIKE "TO BUILD A TEXT SYSTEM THAT CAN READ [LANGUAGE] SO PEOPLE CAN USE PHONES IN [LANGUAGE]"].

Mutumin da ke jagorantar wannan aikin shine [PROJECT LEAD NAME], wanda za ka iya tuntuɓa a [EMAIL / PHONE / WHATSAPP]. Idan kana da damuwa game da wannan aikin wanda ba ka son tattaunawa da [LEAD NAME], za ka iya tuntuɓar mai shiga tsakani (ombudsperson) mai zaman kansa na aikin [OMBUDSPERSON NAME] a [CONTACT]; duba Sashe na D.

Abin da muke buƙata daga gare ka​

Muna roƙonka da ka bayar da gudummawar [DESCRIBE THE CONTRIBUTION, e.g., "one hour of recorded speech in [LANGUAGE]", "one hundred short written sentences in [LANGUAGE]", "labels for 500 social media posts", "translations of 200 sentences into [LANGUAGE]", "consent for images of your handwritten notes to be photographed"].

Gudummawarka za ta:

  • Zama [PAID / UNPAID / VOLUNTEER WITH SMALL COMPENSATION]. Idan da biyan kuɗi: [AMOUNT + CURRENCY + PAYMENT SCHEDULE + PAYMENT METHOD, e.g., mobile money].
  • Ɗauki kusan [TIME ESTIMATE].
  • A yi rakodinta a [WHERE, IN WHAT FORMAT: audio / text / image].
  • Zama [ANONYMISED / NAME-ATTRIBUTED / NAME-OPTIONAL (pick one)].

Abin da za mu yi da gudummawarka​

Za a yi amfani da gudummawarka don:

  1. [SPECIFIC USE ONE, e.g., "train a text-classifier model for Hausa"]
  2. [SPECIFIC USE TWO, e.g., "release publicly as an open dataset so other researchers can also work on Hausa"]
  3. [SPECIFIC USE THREE, e.g., "be included in academic publications with your contribution credited"]

Ba za mu:

  • [SPECIFIC AVOID ONE, e.g., "sell your voice recording to companies without asking you first"]
  • [SPECIFIC AVOID TWO, e.g., "share your name with anyone outside the project without asking you first"]
  • [SPECIFIC AVOID THREE, for voice or biometric data: "use your voice to make it say things you did not actually say"]
  • [SPECIFIC AVOID FOUR, e.g., "allow it to be used for surveillance or to identify you"]

Idan muka yanke shawarar yin amfani da gudummawarka don wani abu da ba ya cikin wannan jerin, za mu tuntuɓe ka don neman izini tukuna.

Yadda za a adana bayananka​

  • Za a adana a: [STORAGE LOCATION: country + institution + system]
  • Za a riƙe na tsawon: [PERIOD, e.g., "5 years, then reviewed for further extension"]
  • Waɗanda za su iya gani sune: [PROJECT TEAM ROLES + external reviewers if any]
  • Idan mun saki bayanan ga jama'a: [LICENCE, e.g., CC BY-NC 4.0] (duba ƙasa).

Saki ga jama'a​

Muna shirin sakin [THE DATA / THE MODEL BUILT FROM IT / BOTH] ga jama'a, a ƙarƙashin lasisin da ke buƙatar duk wanda zai yi amfani da shi ya:

  • [ATTRIBUTE THE COMMUNITY IN A SPECIFIED WAY]
  • [NOT USE IT COMMERCIALLY (IF CC BY-NC) OR ANY OTHER RESTRICTIONS]
  • [OTHER TERMS FROM YOUR LICENCE CHOICE]

Za a rubuta cikakkun sharuɗɗan lasisin a [URL] a ranar da aka saki bayanan.

Idan shawarar al'umma ta canza game da yadda ya kamata a saki bayanan, [NAMED DECISION BODY / STEWARD] yana da ikon sabunta wannan shawarar har zuwa [DATE OF FIRST PUBLIC RELEASE].

Haƙƙoƙinka​

Kana da haƙƙin:

  • Janye gudummawarka a kowane lokaci kafin a saki bayanan ga jama'a, ta hanyar tuntuɓar [NAMED CONTACT + CHANNEL] da kuma faɗin lambar sirrinka ta mai bayar da gudummawa (duba ƙasa).
  • Neman kwafin abin da muka tattara daga gare ka a kowane lokaci.
  • Ƙin bayar da gudummawar wani takamaiman abu, ɗaukar hutu, ko tsayawa a kowane lokaci koda bayan sanya hannu a wannan fom, ba tare da wani hukunci ba. Ba kwa buƙatar bayar da wani dalili.
  • Yin tambayoyi game da aikin, gudummawarka, ko bayananka a kowane lokaci. Ƙungiyar ta ɗauki alƙawarin amsawa a cikin [TIME, e.g., "seven days"].

Bayan an saki bayanan ga jama'a, ba za a iya janye gudummawarka daga kwafin da tuni suke yawo ba. Ba za mu saka gudummawarka a cikin duk wani saki na gaba ba.

Lambar sirrinka ta mai bayar da gudummawa​

Lambar sirrinka ta mai bayar da gudummawa ita ce: [GENERATED CODE, e.g., HAU-A0037]

Ajiye wannan lambar. Za ka buƙace ta don gane kanka idan ka tuntuɓe mu game da gudummawarka.

Ta hanyar sanya hannu a ƙasa, ka tabbatar da cewa:

  • Ka karanta (ko an karanta maka) wannan fom a cikin harshen da kake fahimta.
  • An amsa tambayoyinka.
  • Ka amince da radin kanka, ba tare da matsin lamba ba.
  • Ka fahimci cewa za ka iya janyewa kafin a saki bayanan ga jama'a.
  • Kana da shekaru [AGE OF LEGAL CONSENT IN THIS JURISDICTION] ko fiye.

Sunan mai bayar da gudummawa (a rubuce): __________________________________________

Sa hannu ko tambarin yatsa: ______________________________________________

Kwanan wata: __________________________________________________________________

Sunan shaida (mamban ƙungiyar aiki): ___________________________________

Sa hannun shaida: _____________________________________________________

Shaida mai zaman kansa (ana buƙata don amincewa da baki, an ba da shawarar don a rubuce; duba Sashe na B): ___________________


Tsarin da zai yi aiki don amincewa daga masu bayar da gudummawa waɗanda ba su iya karatu da rubutu ba, masu bayar da gudummawa waɗanda suka fi son amincewa da baki ba tare da la'akari da ilimin boko ba, da kuma yanayin da fom ɗin a rubuce bai dace ba.

Kafin yin rakodi​

  • Dole ne mai yin hira ya kasance ɗan asalin mai magana da harshen mai bayar da gudummawa, ko kuma a samar da fassara ta hanyar wani amintaccen mutum na uku (ba ma'aikacin aikin ba).
  • Dole ne shaida mai zaman kansa ya kasance a wajen. Shaidan wani mamban al'umma ne da mai bayar da gudummawa ya amince da shi, ba mamban ma'aikatan aikin ba. Idan mai bayar da gudummawa ba zai iya kawo shaida ba, sai a sake tsara lokacin hirar. Kada a ci gaba ba tare da shaida ba.
  • Ana ajiye kayan taimakon gani na amincewa (visual consent aid) (a ƙasa) a gaban mai bayar da gudummawa. Wani taƙaitaccen bayani ne mai hoto na shafi ɗaya na abin da bayanan suke, yadda za a yi amfani da su, wanda zai riƙe su, da kuma yadda za a iya janye amincewa. Shirya wannan tun da wuri; ba madadin tattaunawa da baki ba ne amma abokin tafiyarta ne.

Rakodin​

Tattaunawar amincewa da aka yi rakodinta a sauti tana ɗaukar abubuwa iri ɗaya da fom ɗin a rubuce, a cikin ainihin harshen mai bayar da gudummawa, an tsara ta a matsayin tattaunawa ba jawabin mutum ɗaya ba. Tattauna:

  1. Su waye mu kuma me muke yi. Gabatar da kanka, aikin, da mai shiga tsakani (ombudsperson).
  2. Abin da muke buƙata daga gare ka. A cikin kalmomin da mai bayar da gudummawa zai fahimta, ba kalmomin fasaha na aikin ba.
  3. Abin da za mu yi da gudummawarka. Saka takamaiman jerin abubuwan da za a yi da takamaiman jerin abubuwan da ba za a yi ba. Tambayi mai bayar da gudummawa ya bayyana abin da ya fahimta; a nan ne rashin fahimta ke fitowa fili.
  4. Har tsawon wane lokaci, a ina, wa zai iya gani. Takamaiman wurin ajiya da tsawon lokacin da za a riƙe.
  5. Shirye-shiryen saki ga jama'a. Bayyana a fili game da lasisin da kuma abin da hakan ke nufi a aikace ("wannan yana nufin wani a wata ƙasa zai iya amfani da shi don gina tsarin waya, amma ba za su iya sayar da shi ba").
  6. Haƙƙoƙi. Janyewa, kwafi, ƙi, tambayoyi.
  7. Lambar sirrin mai bayar da gudummawa. Faɗi lambar da ƙarfi domin ta shiga cikin rakodin; tambayi mai bayar da gudummawa ya maimaita ta.
  8. Amincewa a fili. "Ka amince ka bayar da gudummawa?", an yi rakodinta da baki.

Ana adana cikakken rakodin sauti tare da bayanan. Shine shaidar amincewa; kula da shi da irin kulawar da ake ba rumbun bayanan (corpus).

Takarda mai girman A4 ko girman wasiƙa ta shafi ɗaya, wacce aka buga, tana nuna:

  • Hoton bayanan da ake tattarawa (makurofon da raƙuman sauti don murya; shafukan rubutu don rubutawa; kyamara don bayanan hoto).
  • Hoton inda zai tafi (kwamfuta, cibiyar adana bayanai, wanda al'ummarka za ta iya ganewa a gani).
  • Hoton wanda zai yi amfani da shi (mutanen da ke amfani da tsarin da aka gina daga bayanan).
  • Hoton da ke wakiltar jerin "ba za mu yi ba": hannu yana tura kuɗin kasuwanci don lasisin da ba na kasuwanci ba; akwati a kulle don ajiya ta sirri; duk abin da zai isar da saƙon ƙuntatawar.
  • Hoto da gajeren bayani don janyewa (katin tuntuɓa mai lambar waya).

Tsara wannan tare da al'umma. Alamomin hotuna da aka tsara don yanayin Turai ko Amurka galibi ba sa isar da saƙo yadda ya kamata. Gwada kayan taimakon gani tare da ƙaramin adadin mambobin al'umma kafin a fara amfani da shi gadan-gadan.


Amincewar mutum ɗaya ba ta ba ka izinin amfani da harshen al'umma don aikin da al'ummar ba ta amince da shi ba. Wannan ya bambanta da kowace amincewar mutum ɗaya da aka samu.

Wacece "al'umma" a nan?​

Don harshen da ake magana da shi a ƙasashe daban-daban da kuma mazauna ƙasashen waje (diaspora), "al'umma" ba wuri ɗaya take ba. Gano su waye:

  • Hukumomin da aka sani: sarakunan gargajiya, dattawa, majalisun al'umma, shugabannin addini, wakilan da aka zaɓa, dangane da tsarin yankin.
  • Ƙungiyoyin masu magana da harshen: ƙungiyoyin kiyaye harshe, ƙungiyoyin al'adu, cibiyoyin ilimi da suka kware a harshen.
  • Wakilan mazauna ƙasashen waje: don harsunan da ke da ɗimbin al'ummomin mazauna ƙasashen waje (diaspora) waɗanda za a saka ko a cire bayanansu.
  • Ƙananan al'ummomi masu rauni: idan tattara bayanan ya shafi wasu takamaiman ƙungiyoyi (mata a wani takamaiman yanki, 'yan gudun hijira, tsiraru a cikin al'umma), wakilansu musamman.

Ba kowane aiki ne ke buƙatar amincewa daga kowane ɗayan waɗannan ba. Hakkin aikin shine gano waɗanda suke da muhimmanci ga tattara bayanan da kuma samun amincewarsu a fili.

An tsara ta a matsayin tuntuɓa a hukumance, an rubuta bayanan taron, tare da amincewa da bayanan taron kuma an raba su ga al'umma don tantancewa.

Tattauna:

  1. Girman aikin, manufa, da al'ummar da za ta amfana: an bayyana a fili kuma da gaskiya.
  2. Wadanne bayanai za a tattara kuma daga wajen wa: nau'o'i, yawa, inda za a samo su.
  3. Tsarin haƙƙin mallakar fasaha (IP): wa ke da me, wane lasisi za a yi amfani da shi wajen sakin bayanan, menene rawar al'umma a cikin shawarwari bayan an saki bayanan. Duba licence selection.
  4. Alƙawuran da aikin ya ɗauka: abin da zai samar, a wane lokaci, da kuma abin da zai faru idan ba a cika alƙawuran ba.
  5. Wakilan al'umma da aka ambata: wanda ke da ikon yin magana a madadin al'umma akan wannan aikin bayan tuntuɓar.
  6. Takamaiman damuwar al'umma: an ba da damar a jera su, tare da martanin aikin ga kowannensu a rubuce a cikin bayanan taron.
  7. Amincewa a fili: rubutacciyar sanarwar amincewar al'umma da wakilan da aka ambata suka sanya hannu a kai.

Ana adana bayanan taron da sanarwar da aka sanya hannu a kai tare da bayanan a matsayin shaidar amincewar al'umma.

Rashin amincewar al'umma​

Idan al'umma ta ƙi amincewa, aikin ba zai ci gaba a kan wannan harshen ba, a cikin wannan tsarin. Wannan kyakkyawan sakamako ne. Kada a yi amfani da "an tuntuɓi al'umma" a matsayin rufa-rufa ga aikin da al'umma ta ƙi.

Idan al'umma ta rabu gida biyu da gaske, yi aiki da tsarin warware rikicin al'ummar kanta kafin a ci gaba. Kada a yi amfani da amincewar wani ɓangare ɗaya a matsayin izini.


Sashe na D: Matsayin mai shiga tsakani (Ombudsperson)​

Mai shiga tsakani (ombudsperson) mutum ne mai zaman kansa daga ƙungiyar aikin wanda masu bayar da gudummawa za su iya kai ƙorafinsu wajensa. Wannan matsayin ya wanzu ne don hana yanayin gazawa inda mai bayar da gudummawa ke jin ba zai iya kai ƙorafi ga ma'aikatan aikin ba, galibi saboda bambancin iko tsakanin mai bincike da mai bayar da gudummawa.

Mai shiga tsakani (ombudsperson):

  • Al'umma ce ke zaɓar sa, ba ƙungiyar aikin ba.
  • Yana da damar tuntuɓar hukumar yanke shawara ta aikin kai tsaye, ba ta hanyar jagoran aikin ba.
  • Ana biyansa diyya don rawar da ya taka a farashi mai adalci na al'umma.
  • Yana da cikakken iko na dakatar da aikin idan wata damuwa tana buƙatar bincike.
  • An ambaci sunansa a cikin fom ɗin amincewa da kuma a cikin kayan taimakon gani, tare da hanyar tuntuɓa wacce ba ta bi ta kan ƙungiyar aikin.

An rubuta matsayin mai shiga tsakani a cikin wata yarjejeniya ta daban a rubuce tsakaninsa da cibiyar da ke ɗaukar nauyin aikin, don haka wa'adin aikinsa ba ya ƙarƙashin ikon ƙungiyar aikin.


Munanan dabi'u (Anti-patterns) da ya kamata a gujewa​

  1. Daidaita fom ɗin a rubuce kawai da tsallake tsarin magana da baki. Ga yawancin ayyukan harsunan Afirka, yawancin masu bayar da gudummawa sun fi gane amincewa da baki.
  2. Fassara fom ɗin zuwa harshen mai bayar da gudummawa da na'urar fassara da tsallake bitar ɗan adam. Fom ɗin amincewa da aka fassara shi ba daidai ba ba amincewa ba ce mai cikakken bayani; wata al'ada ce kawai wacce ba za ta tsallake bita ba.
  3. Yin amfani da ƙungiyar aiki a matsayin "shaida" don amincewa da baki. Dole ne shaidan ya zama mai zaman kansa; ma'aikatan aikin ba masu zaman kansu ba ne.
  4. Tsallake kayan taimakon gani saboda "mai bayar da gudummawa yana jin Ingilishi/Faransanci dama". Kayan taimakon gani ba don ƙwarewar harshe ba ne; don sauƙaƙa fahimtar manufa ne. Wasu manufofin (buɗaɗɗen lasisi, cibiyar adana bayanai, sakin samfuri) ba su da fassara ta kai tsaye a kowane harshe.
  5. Adana shaidar amincewa daban da bayanan. Dole ne su tafi tare. In ba haka ba rumbun bayanan (corpus) zai zama mai yiwuwar amfani ba tare da bayanan amincewa ba, kuma bayanan amincewa zai zama shaidar bincike kawai maimakon wanda ke aiki.
  6. Amincewar al'umma daga "dattijo" ko "shugaba" guda ɗaya a matsayin madadin tuntuɓa a hukumance. Ikon mutum ɗaya ba ɗaya yake da amincewar al'umma ba koda a cikin tsarin da ke da tsauraran matakan shugabanci.
  7. Yin amfani da bayanan taron tuntuɓar al'umma a matsayin takardar tallace-tallace. Bayanan taron wani abin adanawa ne; kiyaye su a kan gaskiya, kuma a magance rashin jituwa a cikin bayanan, ba ta hanyar zaɓar wasu maganganu daga baya ba.

Ƙarin karatu​


Lura ga mai bayar da gudummawa. Idan ka gudanar da ainihin aikin Sarrafa Harshen Jama'a (NLP) na harshen Afirka tare da tsarin amincewa mai aiki (ko kuma idan ka gaji matsalar amincewa daga aikin da ya gabata kuma dole ka sake tsara shi), gudummawa mafi daraja ita ce fasalin wannan samfurin da aka yi wa gyara tare da bayanin abin da ka yi amfani da shi a zahiri, abin da ya yi aiki, da abin da bai yi aiki ba. Ainihin bayanan amincewa daga ainihin ayyuka sune ainihin gaskiyar da wannan samfurin ke ƙoƙarin tunkara.

Join the discussion

Spotted an error, have a question, or want to share what worked on a real project? Sign in with GitHub to add your voice. Every thread lives in the open, powered by GitHub Discussions.

Loading discussion…
Masakhane African Languages Hub logo

Supported by

Masakhane African Languages Hub

The AfriPlaybook is supported by the Masakhane African Languages Hub, a pan-African initiative building open, culturally grounded datasets for African languages. We are grateful for its support.

About the Hub

In collaboration with

  • Bayero University, Kano
  • Bahir Dar University
  • HausaNLP
  • EthioNLP