Respan Dataset Explorer

Select one behavior. Every returned turn has one binary label: Present or Absent. Source: final dense boolean release.

5,167,182physical rows

86shards

0.00%qualified row coverage

0.00%qualified cell coverage

Shard Start offset Rows Behavior Label Random row JSON API

turns-00058.parquet:55990

bef3aebe991d00182ca73c80

turn 1/1gpt-4o-mini-2024-07-18TurkishUnited States426 words

degenerate_repetitionAbsentFinal dense release

USER

Assistant:
User: System: Comment Analysis Prompt for App Store Reviews

You will receive user comments from app store reviews. Follow the step-by-step instructions below to extract key information and analyze the emotional tone of the comment. Provide your response strictly in JSON format, and ensure the output language is Turkish.

Step 1: Keyword Extraction
Identify and extract the most important keywords or phrases from the comment that represent the central themes. Focus on issues, features, requests, or any notable patterns in the user's feedback. Use concise and descriptive keywords.

- Output Format:
Keywords = ["keyword 1", "keyword 2", "keyword 3", ...]

Step 2: Sentiment Analysis
Determine the emotional tone of the comment and classify it into one of the following categories:
- Pozitif (Positive): The comment expresses satisfaction, praise, or a positive attitude.
- Negatif (Negative): The comment reflects dissatisfaction, criticism, or a negative experience.
- Nötr (Neutral): The comment is neutral, factual, or does not convey a strong emotional tone.

Additionally, provide a confidence score for the classification as a decimal value between 0 and 1.

Output Requirements
- Present your analysis in JSON format.
- All outputs must be in Turkish.
- Strictly follow the example structure provided below.

Example Outputs

Example 1:
Given Text:
"Bu uygulama gerçekten harika! Kullanımı çok kolay ve arayüzü çok sezgisel. Müşteri destek ekibi özellikle yardımcı oldu ve hızlı bir şekilde yanıt verdi. Kesinlikle paraya değer. Artık her gün kullanıyorum, kesinlikle tavsiye ederim!"

Response:
{
"action": "Keyword Extraction and Sentiment Analysis",
"keywords": ["Kolay kullanım", "Sezgisel arayüz", "Müşteri desteği", "Hızlı yanıt", "Paraya değer", "Tavsiye"],
"Sentiment Analysis": "Pozitif",
"confidence": 0.95
}

Example 2:
Given Text:
"Uygulama ilk başta iyi gibiydi ama birkaç gün sonra ciddi sorunlar yaşamaya başladım. Uygulama sürekli çöküyor ve açılmıyor. Ayrıca bazı özellikler beklendiği gibi çalışmıyor. Destek ekibiyle iletişime geçmeye çalıştım ama bir yanıt alamadım. Gerçekten hayal kırıklığına uğradım."

Response:
{
"action": "Keyword Extraction and Sentiment Analysis",
"keywords": ["Ciddi sorunlar", "Sürekli çökme", "Açılmama", "Çalışmayan özellikler", "Destek ekibi", "Yanıt yok", "Hayal kırıklığı"],
"Sentiment Analysis": "Negatif",
"confidence": 0.90
}

Instructions to the Model
- Extract concise keywords that capture the essence of the comment.
- Categorize the sentiment as Pozitif, Negatif, or Nötr.
- Ensure the confidence score is a decimal (e.g., 0.90).
- Format the output in JSON as demonstrated.

Final Note: Ensure accuracy and completeness in all responses.
User: Kaliteli ürün Uygun fiyat kibar kuryeler tarafından zamanında teslimat emeği geçen herkese çok teşekkürler
Assistant:
Assistant:

ASSISTANT

{
    "action": "Keyword Extraction and Sentiment Analysis",
    "keywords": ["Kaliteli ürün", "Uygun fiyat", "Kibar kuryeler", "Zamanında teslimat", "Teşekkür"],
    "Sentiment Analysis": "Pozitif",
    "confidence": 0.93
}

turns-00058.parquet:55991

b49f228e68f6dc3c8cf952ec

turn 1/1gpt-4o-mini-2024-07-18TurkishSweden410 words

degenerate_repetitionAbsentFinal dense release

USER

Assistant:
User: System: Comment Analysis Prompt for App Store Reviews

- Output Format:
Keywords = ["keyword 1", "keyword 2", "keyword 3", ...]

Additionally, provide a confidence score for the classification as a decimal value between 0 and 1.

Output Requirements
- Present your analysis in JSON format.
- All outputs must be in Turkish.
- Strictly follow the example structure provided below.

Example Outputs

Final Note: Ensure accuracy and completeness in all responses.
User: Fiyatlar uygun Teşekkürler A101🙏🥰💐
Assistant:
Assistant:

ASSISTANT

{
    "action": "Keyword Extraction and Sentiment Analysis",
    "keywords": ["Uygun fiyatlar", "Teşekkürler"],
    "Sentiment Analysis": "Pozitif",
    "confidence": 0.92
}

turns-00058.parquet:55992

c9d254b147bb237bbffab608

turn 1/1gpt-4o-mini-2024-07-18TurkishUnited Kingdom423 words

degenerate_repetitionAbsentFinal dense release

USER

Assistant:
User: System: Comment Analysis Prompt for App Store Reviews

- Output Format:
Keywords = ["keyword 1", "keyword 2", "keyword 3", ...]

Additionally, provide a confidence score for the classification as a decimal value between 0 and 1.

Output Requirements
- Present your analysis in JSON format.
- All outputs must be in Turkish.
- Strictly follow the example structure provided below.

Example Outputs

Final Note: Ensure accuracy and completeness in all responses.
User: Cok kıbar anlayıslı bır beyefendıydı cok ınceydı tesekkur ederıö evımın tam onune getırdıler mınnettarım
Assistant:
Assistant:

ASSISTANT

{
    "action": "Keyword Extraction and Sentiment Analysis",
    "keywords": ["Kibar", "Anlayışlı", "İnce", "Teşekkür", "Minnettar", "Getirme"],
    "Sentiment Analysis": "Pozitif",
    "confidence": 0.92
}

turns-00058.parquet:55993

a6757fa6877f069beabbfa44

turn 1/1gpt-4o-mini-2024-07-18TurkishMalaysia498 words

degenerate_repetitionAbsentFinal dense release

USER

Assistant:
User: System: Comment Analysis Prompt for App Store Reviews

- Output Format:
Keywords = ["keyword 1", "keyword 2", "keyword 3", ...]

Additionally, provide a confidence score for the classification as a decimal value between 0 and 1.

Output Requirements
- Present your analysis in JSON format.
- All outputs must be in Turkish.
- Strictly follow the example structure provided below.

Example Outputs

Final Note: Ensure accuracy and completeness in all responses.
User: Alışveriş listemi oluşturdum ve saat 12:00 ile 14:30 arası teslimat seçeneğini seçtim. Saat 3’te hala sipariş durumu hazırlanıyor olarak gördüğümde arayıp durumu ilettim. İlgili yere bildiriyorum deyip kapattı. 10 dakika sonra bildirim geldi siparişiniz yola çıktı diye fakat bundan 5 dakika sonra siparişiniz talebinize istinaden iptal edilmiştir diye bildirim geldi. Tekrar aradım cevap olarak söyledikleri; kurye sebep belirtmeden iptal etmiş yapabileceğimiz bir şey yok oldu. Hafta sonu 4 saat beni esir ettiler evde beklemek zorunda kaldım sonuç fiyasko oldu. Sakın A101 uygulamasından alışveriş yapmayın!
Assistant:
Assistant:

ASSISTANT

{
    "action": "Keyword Extraction and Sentiment Analysis",
    "keywords": ["Alışveriş listesi", "Teslimat seçeneği", "Sipariş durumu", "İptal", "Kurye", "Fiyasko", "A101 uygulaması"],
    "Sentiment Analysis": "Negatif",
    "confidence": 0.92
}

turns-00058.parquet:55994

6befb83a7a7e4e93b44f982b

turn 1/1gpt-4o-mini-2024-07-18Turkishunknown country419 words

degenerate_repetitionAbsentFinal dense release

USER

Assistant:
User: System: Comment Analysis Prompt for App Store Reviews

- Output Format:
Keywords = ["keyword 1", "keyword 2", "keyword 3", ...]

Additionally, provide a confidence score for the classification as a decimal value between 0 and 1.

Output Requirements
- Present your analysis in JSON format.
- All outputs must be in Turkish.
- Strictly follow the example structure provided below.

Example Outputs

Final Note: Ensure accuracy and completeness in all responses.
User: Teslim edemiyorlar
Kurye ürünün teslimatını iptal etti diye saçma cevaplar veriyorlar
Assistant:
Assistant:

ASSISTANT

{
    "action": "Keyword Extraction and Sentiment Analysis",
    "keywords": ["Teslimat", "Kurye", "İptal", "Saçma cevaplar"],
    "Sentiment Analysis": "Negatif",
    "confidence": 0.85
}

turns-00058.parquet:55995

6a651852fcfc665aca7ee690

turn 1/1gpt-4o-mini-2024-07-18TurkishUnited States451 words

degenerate_repetitionAbsentFinal dense release

USER

Assistant:
User: System: Comment Analysis Prompt for App Store Reviews

- Output Format:
Keywords = ["keyword 1", "keyword 2", "keyword 3", ...]

Additionally, provide a confidence score for the classification as a decimal value between 0 and 1.

Output Requirements
- Present your analysis in JSON format.
- All outputs must be in Turkish.
- Strictly follow the example structure provided below.

Example Outputs

Final Note: Ensure accuracy and completeness in all responses.
User: Asla seçilen vakit aralığında teslimat yok, profesyonel değil tedarikçi mağazalar ve kuryeler. Teslim edilmeyen ürünlerin ödemesini iade etmekte zorluk çıkarıyorlar ki yol tüketici hakem heyetine gidiyor. Müşteri hizmetleri göreceli daha normal.
Assistant:
Assistant:

ASSISTANT

{
    "action": "Keyword Extraction and Sentiment Analysis",
    "keywords": ["Seçilen vakit aralığı", "Teslimat yok", "Profesyonel değil", "Tedarikçi mağazalar", "Kuryeler", "İade zorluğu", "Tüketici hakem heyeti", "Müşteri hizmetleri"],
    "Sentiment Analysis": "Negatif",
    "confidence": 0.92
}

turns-00058.parquet:55996

f7b4477be633d3c2feb44df5

turn 1/1gpt-4o-mini-2024-07-18TurkishChina422 words

degenerate_repetitionAbsentFinal dense release

USER

Assistant:
User: System: Comment Analysis Prompt for App Store Reviews

- Output Format:
Keywords = ["keyword 1", "keyword 2", "keyword 3", ...]

Additionally, provide a confidence score for the classification as a decimal value between 0 and 1.

Output Requirements
- Present your analysis in JSON format.
- All outputs must be in Turkish.
- Strictly follow the example structure provided below.

Example Outputs

Final Note: Ensure accuracy and completeness in all responses.
User: Teslimat cok guzel bi kere yerimi tarif ettim devamlı sorunsuz sipariş geçebiliyorum .teşekkürler
Assistant:
Assistant:

ASSISTANT

{
    "action": "Keyword Extraction and Sentiment Analysis",
    "keywords": ["Teslimat", "Yer tarifü", "Sorunsuz sipariş", "Teşekkürler"],
    "Sentiment Analysis": "Pozitif",
    "confidence": 0.93
}

turns-00058.parquet:55997

fd098864e7b948d00ef5cecd

turn 1/1gpt-4o-mini-2024-07-18TurkishMexico406 words

degenerate_repetitionAbsentFinal dense release

USER

Assistant:
User: System: Comment Analysis Prompt for App Store Reviews

- Output Format:
Keywords = ["keyword 1", "keyword 2", "keyword 3", ...]

Additionally, provide a confidence score for the classification as a decimal value between 0 and 1.

Output Requirements
- Present your analysis in JSON format.
- All outputs must be in Turkish.
- Strictly follow the example structure provided below.

Example Outputs

Final Note: Ensure accuracy and completeness in all responses.
User: Harika 👍🏻
Assistant:
Assistant:

ASSISTANT

{
    "action": "Keyword Extraction and Sentiment Analysis",
    "keywords": ["Harika"],
    "Sentiment Analysis": "Pozitif",
    "confidence": 0.98
}

turns-00058.parquet:55998

63b0613fbb121c18d4e283ec

turn 1/1gpt-4o-mini-2024-07-18TurkishUnited States446 words

degenerate_repetitionAbsentFinal dense release

USER

Assistant:
User: System: Comment Analysis Prompt for App Store Reviews

- Output Format:
Keywords = ["keyword 1", "keyword 2", "keyword 3", ...]

Additionally, provide a confidence score for the classification as a decimal value between 0 and 1.

Output Requirements
- Present your analysis in JSON format.
- All outputs must be in Turkish.
- Strictly follow the example structure provided below.

Example Outputs

Final Note: Ensure accuracy and completeness in all responses.
User: Tamamen yanlış ütünler teslim edildi. Canlı detsek de hiçbir şekilde sorun çözmüyor trendyol getir vb uygulamalardaki düzeni dakikliği beklemeyin berbat ötesi kötü. Teslimat çok geç yetmiyor gibi yanlış alakasız ürünler gönderiyorlar.
Assistant:
Assistant:

ASSISTANT

{
    "action": "Keyword Extraction and Sentiment Analysis",
    "keywords": ["Yanlış ürünler", "Canlı destek", "Sorun çözme", "Teslimat gecikmesi", "Alakasız ürünler", "Berbat deneyim"],
    "Sentiment Analysis": "Negatif",
    "confidence": 0.92
}

turns-00058.parquet:55999

bee3b729bc6ee649e2079fc1

turn 1/1gpt-4o-mini-2024-07-18TurkishLithuania419 words

degenerate_repetitionAbsentFinal dense release

USER

Assistant:
User: System: Comment Analysis Prompt for App Store Reviews

- Output Format:
Keywords = ["keyword 1", "keyword 2", "keyword 3", ...]

Additionally, provide a confidence score for the classification as a decimal value between 0 and 1.

Output Requirements
- Present your analysis in JSON format.
- All outputs must be in Turkish.
- Strictly follow the example structure provided below.

Example Outputs

Final Note: Ensure accuracy and completeness in all responses.
User: Güzel servis yaklaşık 1 saat içinde geldi eksiksiz şekilde tşk
Assistant:
Assistant:

ASSISTANT

{
    "action": "Keyword Extraction and Sentiment Analysis",
    "keywords": ["Güzel servis", "Hızlı teslimat", "Eksiksiz", "Teşekkür"],
    "Sentiment Analysis": "Pozitif",
    "confidence": 0.98
}