Teburin Abubuwan Ciki
1. Gabatarwa
Canjin dijital na koyon harshen Sinanci ya ƙaru sosai a lokacin annobar COVID-19, tare da Cibiyoyin Confucius suna canzawa zuwa dandamali na kan layi kuma suna aiwatar da Shirye-shiryen Aiki na 2021-2025 don Ilimin Sinanci na Duniya. Wannan bita na tsari yana binciken karatu 29 daga 2017-2022 da suka mayar da hankali kan wasannin ilmantarwa da Tsare-tsaren Koyarwa na Hankali (ITS) a cikin koyon harshen Sinanci.
Nazarin Karatu 29
Cikakken bita na binciken kwanan nan
2017-2022
Lokacin wallafa da aka rufe
Rukunan Fasaha 3
Wasanni, Amfani da Wasanni, da ITS
2. Hanyar Bincike
2.1 Dabarun Bincike
Binciken na tsari ya yi amfani da bincike mai tsauri na bayanai a duk faɗin ScienceDirect da Scopus, ta amfani da kalmomin da suka haɗa da "koyon harshen Sinanci," "wasannin ilmantarwa," "tsare-tsaren koyarwa na hankali," da "hankalin wucin gadi." An iyakance binciken zuwa wallafe-wallafen takwarorinsu daga 2017 zuwa 2022 don ɗaukar sabbin ci gaban fasaha.
2.2 Ma'aunin Haɗawa
An haɗa karatu bisa takamaiman ma'auni: bincike na zahiri da ke mai da hankali kan koyon harshen Sinanci da fasaha ta inganta, bayanin hanyar aiki bayyananne, da sakamako masu aunawa masu alaƙa da tasirin koyo, ƙarfafawa, ko gamsuwa. Ma'aunin cirewa ya kawar da takardun ka'idoji ba tare da bayanan zahiri ba da kuma karatun da ba su magance takamaiman koyon harshen Sinanci ba.
2.3 Nazarin Bayanai
Nazarin ya yi amfani da hanyoyi na ƙididdigewa da na inganci, yana binciken girman tasiri daga sakamakon gwaji na farko da na baya, yayin da kuma ake gudanar da nazarin jigo na martani na inganci daga masu koyo da malamai.
3. Sakamako
3.1 Wasannin Ilimantawa
Wasannin ilmantarwa sun nuna tasiri mai mahimmanci akan samun ƙamus da gane haruffa. Karatu sun nuna matsakaicin ƙimar haɓaka 23-35% a cikin riƙewar haruffa idan aka kwatanta da hanyoyin gargajiya. Mafi kyawun wasannin sun haɗa da algorithms na maimaitawa da sikelin wahala mai daidaitawa.
3.2 Tsare-tsaren Koyarwa na Hankali
Aiwatar da ITS ya nuna ƙarfi na musamman a cikin hanyoyin koyo na keɓance da martani na ainihi. Tsarin da suka haɗa da sarrafa harshe na halitta sun sami daidaiton kashi 89% a cikin gane sauti kuma suna ba da gyaran gyara nan take, yana haɓaka ƙwarewar furuci sosai.
3.3 Dabarun Amfani da Wasanni
Abubuwan amfani da wasanni da suka haɗa da maki, lambobin yabo, da jerin gwanon shugabanni sun ƙara haɗakar masu koyo da kashi 42% kuma sun dore yawan shiga. Mafi nasarar aiwatar da su sun daidaita abubuwan gasa tare da fasalolin koyo na haɗin gwiwa.
Mahimman Fahimta
- Koyon da fasaha ta inganta yana haɓaka ƙarfafawa da kashi 67% idan aka kwatanta da hanyoyin gargajiya
- An lura da haɓakar kwarin gwiwar kai a cikin kashi 78% na mahalartan binciken
- Makin gamsuwar koyo ya ƙaru da maki 2.3 akan ma'auni maki 5
- Tsare-tsaren daidaitawa suna nuna ƙimar riƙewa mai kyau da kashi 45% fiye da abun ciki mai tsayi
4. Tattaunawa
4.1 Nazarin Tasiri
Bitar ya nuna bayyananniyar tasirin hanyoyin da fasaha ta inganta, tare da girman tasiri daga d=0.45 zuwa d=0.78 a cikin sakamakon koyo daban-daban. An lura da mafi mahimmancin ci gaba a cikin samun ƙamus da daidaiton furuci.
4.2 Aiwar da Fasaha
Tushen Lissafi
Algorithms na koyo masu daidaitawa a cikin nasarar aiwatar da ITS sau da yawa suna amfani da gano ilimin Bayesian, wanda aka wakilta ta:
$P(L_{n+1}) = P(L_n) + (1 - P(L_n)) imes P(T) imes P(G)$
Inda $P(L_n)$ ke wakiltar yuwuwar sanin fasaha a mataki na n, $P(T)$ shine yuwuwar canjawa, kuma $P(G)$ shine yuwuwar zato.
Misalin Aiwar da Lamba
class AdaptiveChineseTutor:
def __init__(self):
self.student_model = {}
self.knowledge_components = {}
def update_student_model(self, student_id, skill, performance):
"""Sabunta ilimin ɗalibi bisa aiki"""
current_knowledge = self.student_model.get(student_id, {}).get(skill, 0.3)
# Sabunta ilimin Bayesian
if performance > 0.7: # Aiki mai kyau
new_knowledge = current_knowledge + (1 - current_knowledge) * 0.3
else: # Aiki mara kyau
new_knowledge = current_knowledge * 0.8
if student_id not in self.student_model:
self.student_model[student_id] = {}
self.student_model[student_id][skill] = min(new_knowledge, 0.95)
def recommend_content(self, student_id):
"""Ba da shawarar abun cikin koyo bisa samfurin ɗalibi"""
student_skills = self.student_model.get(student_id, {})
weakest_skill = min(student_skills, key=student_skills.get)
return self.select_content(weakest_skill)4.3 Nazari na Asali
Nazarin Kwararre: Fasaha a cikin Ilimin Harshen Sinanci
Gaskiya A Kai Tsaye: Wannan bita ya fallasa babban gibi tsakanin yuwuwar fasaha da aiwatar da koyarwa a cikin ilimin harshen Sinanci. Duk da yake karatun sun nuna sakamako masu ban sha'awa, fannin yana fama da rarrabuwar ci gaba da rashin isasshen haɗin kai tare da ingantattun ka'idodin samun harshe.
Sarkar Hankali: Ci gaban yana bayyananne: dijital da annoba ta haifar → ƙaruwar amfani da wasanni da ITS → ingantattun haɓaka a cikin ƙarfafawa da kwarin gwiwar kai → amma ƙarancin fahimtar mafi kyawun dabarun aiwatarwa. Hanyar da ta ɓace ita ce haɗin kai na tsarin waɗannan fasahohin cikin cikakken ƙirar manhaja, kama da yadda CycleCAN ya kawo juyin juya hali ga fassarar hoto zuwa hoto ta hanyar kafa ingantattun tsare-tsaren canji (Zhu et al., 2017).
Abubuwan Haske da Kaskantarwa: Nasarar da ta fito ita ce haɓakar haɗakar da kashi 42% daga amfani da wasanni - wannan ba kawai ci gaba ne kawai ba, yana da canji. Duk da haka, kaskantarwa tana da tsananin gaske: yawancin karatu suna mai da hankali kan ma'auni na ɗan gajeren lokaci ba tare da magance riƙewa na dogon lokaci ko haɓaka ƙwarewar al'adu ba. Idan aka kwatanta da ingantattun dandamali kamar Duolingo ko hanyoyin da ke da goyan baya na bincike a cikin tsarin Koyarwa na Fahimi na Carnegie Mellon, aiwatar da takamaiman Sinanci ba su da ingantaccen gwajin A/B da ingantaccen inganci wanda zai sa su zama masu jan hankali da gaske.
Gargaɗin Aiki: Hanyar gaba tana buƙatar ƙungiyoyi uku na dabarun: Na farko, karɓi hanyoyin koyo masu canzawa daga ingantattun dandamali na koyon harshen Turanci. Na biyu, haɗa AI mai fahimtar yanayi kama da binciken Lissafin Tasiri daga Dakin MIT Media. Na uku, kafa ma'auni na kimantawa waɗanda suka wuce makin gwaji nan take don auna ainihin ƙwarewar harshe da fahimtar al'adu. Gaskiyar damar ba ta cikin ƙirƙirar ƙarin wasanni ba, amma a cikin gina tsare-tsare masu daidaitawa waɗanda suka fahimci ƙalubalen musamman na samun harshe mai sauti da haddace haruffa - ƙalubalen da ke buƙatar keɓantaccen maganin fasaha fiye da abin da dandamali na koyon harshe na gaba ɗaya ke bayarwa.
Binciken zai amfana da haɗa samfuran gano ilimi kama da waɗanda ake amfani da su a cikin binciken tsare-tsaren koyarwa na hankali a Jami'ar Carnegie Mellon, yayin da kuma ake magance al'amarin al'adu na koyon harshe wanda ya wuce samun ƙamus kawai. Kamar yadda nasarar gine-ginen masu canzawa a cikin sarrafa harshe na halitta (Vaswani et al., 2017) ya nuna, ci gaba na gaba a fasahar harshen Sinanci zai yiwu ya zo daga daidaita waɗannan ingantattun gine-ginen AI musamman don sarrafa harshe mai sauti da ingantaccen koyon haruffa.
Sakamakon Gwaji da Zane-zane
Karatun da aka yi bita akai-akai sun nuna gagarumin ribar koyo. A cikin wani bincike na wakilci, masu koyo da ke amfani da ITS don samun sauti sun nuna:
- Ingantaccen daidaiton gane sauti da kashi 45%
- Rage lokacin koyo da kashi 32% idan aka kwatanta da hanyoyin gargajiya
- Makin gamsuwa mafi girma da kashi 78%
Bayanin Zane: Taswira mai kwatancen sandar zai nuna makin gwaji na farko da na baya a cikin ƙungiyoyi uku: koyarwa ta gargajiya, koyo na tushen wasa, da koyo na taimakon ITS. Ƙungiyar ITS za ta nuna mafi girman makin gwaji na baya, musamman a cikin gwajin furuci da gane haruffa. Wani jadawalin layi na biyu zai nuna lanƙwan koyo, yana nuna ƙungiyar ITS ta cimma ma'auni na ƙwarewa a cikin kusan 30% ƙaramin lokaci.
5. Hanyoyin Gaba
Bitar ta gano hanyoyin bincike masu ban sha'awa da yawa:
5.1 Keɓancewa Mai Ƙarfin AI
Tsare-tsaren gaba yakamata su haɗa da ƙwararrun algorithms na AI don hanyoyin koyo na keɓance, mai yuwuwa ta amfani da gine-ginen masu canzawa kama da samfuran GPT amma an inganta su don koyar da harshen Sinanci.
5.2 Haɗin Koyo Mai Nau'i Daban-daban
Haɗa gane haruffa na gani tare da horon sauti na ji da aikin rubutu ta hannu ta hanyar fasahar tawada ta dijital zai iya haifar da ƙarin gogewar koyo cikakke.
5.3 Aiwatar da Al'adu Daban-daban
Bincike yakamata ya binciko yadda za a iya daidaita waɗannan fasahohin yadda ya kamata don yanayin al'adu daban-daban da salon koyo a cikin masu koyo na duniya.
5.4 Nazarin Tasirin Dogon Lokaci
Bincike na gaba yana buƙatar bincika riƙewar dogon lokaci da aikace-aikacen ainihin duniya na ƙwarewar harshe da aka samu ta hanyar shisshigin fasaha.
6. Bayanan da aka ambata
Hung, H. T., Yang, J. C., Hwang, G. J., Chu, H. C., & Wang, C. C. (2018). A scoping review of research on digital game-based language learning. Computers & Education, 126, 89-104.
Lai, J. W., & Bower, M. (2019). How is the use of technology in education evaluated? A systematic review. Computers & Education, 133, 27-42.
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., ... & Polosukhin, I. (2017). Attention is all you need. Advances in neural information processing systems, 30.
Zhu, J. Y., Park, T., Isola, P., & Efros, A. A. (2017). Unpaired image-to-image translation using cycle-consistent adversarial networks. Proceedings of the IEEE international conference on computer vision, 2223-2232.
Maksimova, A. (2021). Digital transformation in Chinese language education. Journal of Educational Technology Research, 45(3), 234-256.
Confucius Institute Headquarters. (2020). Annual Development Report of Confucius Institutes.