Hanyar bincike
Daga rahotannin tushe zuwa rubutu, jigogi da shaidar da za ku bincika.
Python + WordPress API: tattara rahotanni
Python da WordPress API sun sauke rahotannin jama’a daga shafin ofishin Isa kuma sun adana hanyoyin zuwa kowace majiya.
Beautiful Soup: tsaftace rubutu
Beautiful Soup ya cire lambar gina shafin yanar gizo domin mu iya karantawa da nazarin rubutun rahoton.
Duba majiyoyi: guje wa maimaituwa da jimillar da ba daidai ba
Mun cire kwafe iri ɗaya kuma mun duba lambobi da rahotannin tushensu, tare da ware tsare-tsare daga abin da aka ruwaito an bayar da guje wa jimillar da ke ƙirga masu amfana fiye da sau ɗaya.
TF-IDF + NMF: gano jigogi
scikit-learn yana amfani da TF-IDF don nuna kalmomin da ke bambanta rubuce-rubuce da NMF don gano jigogi 8 masu maimaituwa a zaɓaɓɓun rahotannin Turanci.
Rukunonin da aka zaɓa da hannu: sanya wa ayyuka suna
Mun sanya ayyuka cikin rukunoni ta hanyar karanta rahotannin; aiki ɗaya na iya shiga rukunoni da yawa.
Ƙirgar jimloli: gina girgijen kalmomi
Shirin yana ƙirga kowace jimla sau ɗaya a kowane rahoto don girgijen ra’ayoyi; kallon rukunoni yana ƙirga ayyuka. Manyan sunaye suna nuna ƙarin bayanai.
Majiyoyi da tattaunawa mai jagora: bayyana sakamako
Allon taƙaitaccen bayani yana haɗa lambobi da alaƙoƙi zuwa rahotanni; chatbot yana bin tambayoyin da aka shirya. Rahotannin ba su tabbatar da tasiri ta hanya mai zaman kanta ba, kuma ba su ƙunshi duk tarihin aikinsa ba.
Yadda wakilan AI suka taimaka wa aikinmu
Wanda ya ƙirƙira shafin ya yi aiki tare da wakilan AI na rubuta lambar manhaja don mayar da ra’ayin zuwa shafi mai aiki. Sun taimaka wajen rubuta shirye-shiryen tattara bayanai da nazari, gina shafuka da tambayoyin jagora, shirya fassarorin gwaji da inganta zane bisa ra’ayoyi. Ayyukan wakilai na musamman sun kuma taimaka wajen bincika zaɓuɓɓukan bincike da duba sassan aikin.
Mai aikin ya tsara alkibla kuma ya gyara kalmomi da yadda ake gabatarwa. Wakilan sun gudanar da gwaje-gwajen atomatik kan bayanai, lissafi da aiki a burauza, tare da kiyaye rahotannin tushe kusa da ikirari. Wannan tallafi ba ya maye gurbin hujjar asali ko tabbatar da ingancin fassara. Chatbot na masu ziyara yana bin tambayoyi da amsoshin da aka shirya; ba ya kiran wakilin AI kai tsaye.
Saitunan fasaha
Saitunan tsarin nazarin zaɓaɓɓun rubutun Turanci. An bar sunayen saitunan fasaha da Turanci.
- Features
- TF-IDF; unigrams + bigrams; min_df=3; max_df=0.85; sublinear_tf=True
- Model
- NMF; 8 components; nndsvda; coordinate descent; random_state=42
- Convergence
- max_iter=600; tol=0.0001; selected fit: 839 features, 117 iterations
Tushe da aiwatarwa
Duba yadda aka aiwatar da bayanan hanyar bincike da aka adana:
