Your task is to detect the top 3 topic with their corresponding sentiment in the text…
- User
Your task is to detect the top 3 topic with their corresponding sentiment in the text given to you by the user. The input from the user written in German. Your output should be in English. The format of your reply should be JSON and it should be a list of objects with exactly two keys: topic and sentiment, as follows: In: "Ich bin ganz zufrieden mit der Geschwindigkeit und meine Tarrif ist auch OK. Aber ich habe letzten Monat eine Rechnung bekommen, die viel zu hoch ist. Das ist wirklich ärgerlich."
Out: [ { "topic": "high bill", "sentiment": "negative" }, { "topic": "speed", "sentiment": "positive" }, { "topic": "tariff", "sentiment": "neutral" } ] In: Die Kundenberatung ist sehr schlecht. Es duaert ewig bis man mit jemandem sprechen kann. Ich bin aber mit meinem Handytariff sehr zufrieden. \n Out:
- Assistant
{ "topic_sentiments": [ { "topic": "customer service", "sentiment": "negative" }, { "topic": "waiting time", "sentiment": "negative" }, { "topic": "mobile tariff", "sentiment": "positive" } ] }
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.