CHOP: Haɗa ChatGPT cikin Aikin Gabatar da Magana na EFL - Bincike da Fahimta

Teburin Abubuwan Ciki

1. 1. Gabatarwa & Bayyani
2. 2. Dandamalin CHOP: Ƙira & Aiki
3. 3. Hanyoyin Bincike & Kimantawa
4. 4. Sakamako & Muhimman Bincike
5. 5. Tsarin Fasaha & Bincike
6. 6. Ayyuka na Gaba & Ci Gaba
7. 7. Nassoshi
8. 8. Ra'ayin Mai Bincike: Cikakkiyar Fahimta, Tsarin Ma'ana, Ƙarfafawa & Kurakurai, Fahimta Mai Amfani

1. Gabatarwa & Bayyani

Wannan takarda tana binciken takardar bincike "CHOP: Haɗa ChatGPT cikin Aikin Gabatar da Magana na EFL." Binciken ya magance wata ƙalubale mai mahimmanci a cikin ilimin Turanci a matsayin Harshen Waje (EFL): wahalar da ɗalibai ke fuskanta wajen haɓaka ƙwarewar gabatar da magana mai inganci saboda ƙarancin damar yin aiki da kuma rashin isasshen ra'ayi na musamman. Takardar ta gabatar da CHOP (Dandamali mai hulɗa da ChatGPT don aikin gabatar da magana), wani sabon tsari da aka ƙera don ba da ra'ayi na ainihi, mai ƙarfin AI yayin atisayen gabatarwa.

2. Dandamalin CHOP: Ƙira & Aiki

CHOP dandamali ne na yanar gizo wanda ya haɗa API na ChatGPT don zama kocin gabatarwa na zamani. Babban tsarin aikinsa, kamar yadda aka nuna a Hoto na 1 na PDF, ya ƙunshi:

• Rikodi & Rarraba: ɗalibai suna rikodin atisayen gabatarwarsu yayin da suke bi ta cikin nunin faifai. Dandamalin yana ba da damar yin aikin kowane yanki na musamman.
• Maimaita Sauti & Rubutun Magana: ɗalibai za su iya maimaita sautinsu. Tsarin yana rubuta maganar don bincike.
• Samar da Ra'ayi na AI: A kan buƙata, ChatGPT yana bincika rubutun kuma yana ba da ra'ayi mai tsari bisa ga ƙa'idodin da aka ƙayyade (misali, tsarin abubuwan ciki, amfani da harshe, isarwa).
• Madaidaicin Hulɗa: ɗalibai suna ƙididdige ra'ayin (ma'aunin Likert mai maki 7), suna bitar bayanansu, kuma suna iya yin tambayoyin biyo baya ga ChatGPT don bayani ko zurfin fahimta.

An ƙera shi a fili don dalibi ya zama cibiyar sa, da nufin ƙirƙirar muhallin aiki mai aminci, mai fa'ida.

3. Hanyoyin Bincike & Kimantawa

Binciken ya yi amfani da hanyar haɗaɗɗun hanyoyi:

• Mataki na Farko: Hira da ƙungiyar mai da hankali tare da ɗaliban EFL 5 don gano buƙatu da abubuwan da ake so.
• Gwajin Dandamali: ɗaliban EFL 13 sun yi amfani da dandamalin CHOP don aikin gabatarwarsu.
• Tattara Bayanai:
  • Rajistan hulɗar ɗalibi-ChatGPT.
  • Binciken bayan amfani da abokin ciniki da ra'ayoyi.
  • Kimantawar ƙwararru game da ingancin ra'ayin da ChatGPT ya samar.

Kimantawar ta mai da hankali kan ingancin ra'ayi, yuwuwar koyo, da karɓar mai amfani.

4. Sakamako & Muhimman Bincike

Binciken bayanan da aka tattara ya bayyana wasu muhimman fahimta:

• Ingancin Ra'ayi: ChatGPT ya ba da ra'ayi mai amfani gabaɗaya akan tsarin abubuwan ciki da harshe (nahawu, ƙamus), amma ya nuna iyakancewa wajen kimanta abubuwan da suka shafi isarwa kamar sautin murya, sauri, da yanayin jiki—wuraren da ƙwararrun mutane suka fi ƙwarewa.
• Fahimtar ɗalibi: Mahalarta sun daraja gaggawar da samun damar ra'ayi. Ikon yin aiki a asirce ya rage damuwa. Siffar tambaya da amsa mai hulɗa ta sami yabo musamman don zurfin fahimta.
• Abubuwan Ƙira: Bayyanannen abubuwan da ke haifar da ra'ayi, tsarin tsarin ƙima, da jagorar UI don ingantattun tambayoyin biyo baya an gano su a matsayin muhimman abubuwan da ke tasiri ga gabaɗayan ƙwarewar koyo.
• Raunin da aka Gano: Dogaro da yawa akan rubutun rubutu ya yi watsi da siffofi na gefen harshe. Ra'ayi na iya zama na gabaɗaya ko kuma ya rasa manufofin da suka dace da mahallin.

5. Tsarin Fasaha & Bincike

5.1. Babban Bututun AI

Ƙwararren baya na fasaha na CHOP ya ƙunshi bututu mai bi da bi: Shigar da Sauti → Magana zuwa Rubutu (STT) → Sarrafa Rubutu → Ƙarfafa LLM (ChatGPT) → Samar da Ra'ayi. Ingantaccen aiki ya dogara ne akan injiniyan gaggawa don ChatGPT. Siffa mai sauƙi na ma'anar ƙimar ra'ayi za a iya tunaninta a matsayin jimlar ma'auni:

$S_{feedback} = \sum_{i=1}^{n} w_i \cdot f_i(T)$

Inda $S_{feedback}$ shine jimlar ƙimar ra'ayi don ma'auni, $w_i$ yana wakiltar ma'auni don siffa ta $i$, $T$ shine rubutun da aka rubuta, kuma $f_i(T)$ aiki ne (wanda LLM ke aiwatarwa) wanda ke kimanta rubutun don wannan siffa (misali, masu haɗa ma'ana, amfani da kalmar mahimmanci). Dandamalin yana yiwuwa yana amfani da samfuri na gaggawa mai juyawa wanda ya haɗa da rubutun ɗalibin, abubuwan da ke cikin nunin faifai, da takamaiman ƙa'idodin kimantawa.

5.2. Misalin Tsarin Bincike (Ba Code ba)

Yi la'akari da tsarin bincike don kimanta tsarin ra'ayi na AI kamar CHOP, wanda aka daidaita daga Ƙirar Kimantawar Horarwa ta Kirkpatrick:

1. Martani: Auna gamsuwar mai amfani da amfanin da ake ganin yana da amfani (ta hanyar bincike/ma'aunin Likert).
2. Koyo: Kimanta samun ilimi/ƙwarewa (misali, gwaji kafin/bayan akan ƙa'idodin gabatarwa).
3. Hali: Lura da canja wurin ƙwarewa zuwa gabatarwa na ainihi (kimantawar ƙwararru na gabatarwa na ƙarshe).
4. Sakamako: Kimanta tasirin dogon lokaci (misali, makin darasi, ma'auni na amincewa akan lokaci).

Binciken CHOP ya fi mayar da hankali kan Matakai 1 da 2, tare da kimantawar ƙwararru ta taɓa Mataki na 3.

6. Ayyuka na Gaba & Ci Gaba

Takardar ta ba da shawarar wasu hanyoyi masu ban sha'awa:

• Haɗin Nau'ikan Hanyoyin Sadarwa: Haɗa binciken bidiyo don ba da ra'ayi kan yanayin jiki, kallon ido, da motsin hannu, wucewa daga binciken rubutu kawai. Bincike a cikin AI mai nau'ikan hanyoyin sadarwa, kamar samfuran da ke haɗa sigina na gani da na ji, yana da alaƙa sosai a nan.
• Koyo na Musamman Mai Daidaitawa: Haɓaka algorithms waɗanda ke bin ci gaban mai koyo akan lokaci kuma su daidaita wahalar ra'ayi da wuraren da ake mayar da hankali, kama da dandamalin koyo mai daidaitawa a wasu fagage.
• Haɗawa da LMS na Cibiya: Saka kayan aiki kamar CHOP cikin Tsarin Gudanar da Koyo (misali, Canvas, Moodle) don haɗin kai na tsarin karatu.
• Takamaiman LLM Fine-tuning: Daidaita buɗaɗɗen LLMs (misali, LLaMA, BLOOM) akan ingantattun tarin ra'ayi na gabatarwa da kayan koyarwa na EFL don ƙirƙirar kociyoyi masu takamaiman yanki da inganci.
• Bita na Ƙungiya & Siffofi na Haɗin Kai: Ƙara ayyuka don zaman ra'ayi na ƙungiyoyi ta hanyar AI, haɓaka muhallin koyo na haɗin gwiwa.

7. Nassoshi

1. Cha, J., Han, J., Yoo, H., & Oh, A. (2024). CHOP: Integrating ChatGPT into EFL Oral Presentation Practice. arXiv preprint arXiv:2407.07393.
2. Brown, T., et al. (2020). Language Models are Few-Shot Learners. Advances in Neural Information Processing Systems, 33.
3. Hwang, G.-J., Xie, H., Wah, B. W., & Gašević, D. (2020). Vision, challenges, roles and research issues of Artificial Intelligence in Education. Computers and Education: Artificial Intelligence, 1, 100001.
4. 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 (ICCV). (CycleGAN a matsayin misali na samfuran haɓakawa masu canzawa).
5. OpenAI. (2023). GPT-4 Technical Report. OpenAI. An samo daga https://cdn.openai.com/papers/gpt-4.pdf

8. Ra'ayin Mai Bincike: Cikakkiyar Fahimta, Tsarin Ma'ana, Ƙarfafawa & Kurakurai, Fahimta Mai Amfani

Cikakkiyar Fahimta: CHOP ba wani kocin AI ba ne kawai; yana juyawa daga isar da abubuwan ciki zuwa ƙirƙirar aiki. Ainihin ƙirƙira yana cikin yunƙurinsa na sarrafa mafi girman ɓangaren horon gabatarwa mai cike da albarkatu: madaidaicin ra'ayi na musamman. Wannan yana magance babban matsalar fa'ida a cikin ilimin EFL. Duk da haka, halittarsa ta yanzu tana da iyaka ta hanyar ra'ayinta na duniya mai mayar da hankali kan rubutu, yana ɗaukar gabatarwa a matsayin rubutu maimakon aiki mai nau'ikan hanyoyin sadarwa.

Tsarin Ma'ana: Ma'anar binciken yana da inganci—gano matsala mai raɗaɗi, mai fa'ida (rashin ra'ayi), amfani da fasaha mai rushewa (LLMs), da gina samfuri mafi ƙarancin inganci (CHOP) don gwada ainihin hasashe. Matsi daga ƙungiyoyin mai da hankali zuwa ƙaramin binciken inganci yana bin mafi kyawun ayyuka a cikin binciken EdTech. Kuskuren ma'ana, duk da haka, shine zato na ɓoye cwarewar ChatGPT a cikin samar da rubutu yana canzawa kai tsaye zuwa ƙwarewar koyarwa. Binciken ya gano wannan tazara daidai, amma ainihin gine-ginen har yanzu yana ɗaukar LLM a matsayin baƙar fata maimakon wani ɓangare a cikin tsarin da aka ƙera don koyarwa.

Ƙarfafawa & Kurakurai: Ƙarfin dandamalin shine sauƙinsa mai kyau da amfanin gaggawa. Yana ba da muhallin aiki mara nauyi, wanda shine zinariya ga masu koyo masu damuwa. Siffar tambaya da amsa mai hulɗa hanya ce mai wayo don yaƙar rashin aiki wanda sau da yawa yana addabar kayan aikin AI. Kuskuren da ya haifar, kamar yadda marubutan suka lura, shine gibin nau'in hanyar sadarwa. Ta hanyar yin watsi da sautin murya, sauri, da isarwa ta gani, CHOP yana haɗarin ƙirƙirar masu magana masu gogewa amma masu yuwuwar mutum-mutumi. Kamar horar da mawaƙin piano ta hanyar kimanta takardan waƙar da suke kunna kawai, ba sautin da suke samarwa ba. Bugu da ƙari, ingancin ra'ayi yana da alaƙa da ɓacin rai na abubuwan da GPT ke fitarwa, waɗanda zasu iya zama masu sabani ko kuma su rasa manufofin koyo masu zurfi.

Fahimta Mai Amfani: Ga malamai da masu haɓakawa, hanyar gaba a bayyane take. Na farko, dakatar da ɗaukar wannan a matsayin matsalar NLP kawai. CHOP na gaba dole ne ya haɗa samfuran nau'ikan hanyoyin sadarwa masu sauƙi (tunanin wav2vec don binciken magana, OpenPose don matsayi) don ba da cikakken ra'ayi. Na biyu, karɓi ƙirar "mutum-a-cikin-madauri" tun daga farko. Dandamalin yakamata ya yi alama ga wuraren da ba a sani ba don bitar malami kuma ya koya daga gyare-gyaren ƙwararru, yana haɓaka ƙa'idodinsa a hankali. Na uku, mayar da hankali kan AI mai bayyanawa. Maimakon kawai ba da ra'ayi, tsarin yakamata ya bayyana *dalilin* da yasa aka ba da shawarar (misali, "Yin dakata a nan yana inganta fahimta saboda..."), yana mai da kayan aikin zuwa abokin tarayya na fahimi na gaskiya. A ƙarshe, tsarin kasuwancin bai kamata ya zama sayar da dandamalin ba, amma sayar da fahimta—tarattun bayanai marasa suna akan matsalolin ɗalibai na gama gari waɗanda zasu iya sanar da ƙirar tsarin karatu a matakin cibiya.