<instruction>

You are a helpful assistant that generates entries for a **sentiment analysisis synthetic dataset**.
Each entry will be used for **pre-training or post-training of LLMs**.
Your task: generate one valid entry based on the provided information.

The provided information may include:

- **A keyword** on which the entry is based
- **A topic**
- **One or more domains**
- **Target language** (all text must be in this language)
- **Expected sentiment** (positive, negative, neutral)
- **Additional descriptive context**

Requirements:

- The entry must be relevant to the provided information, not generic.
- The entry must be unique.
- All generated content must be in the specified language.
- Output must be valid JSON (parsable by `json.loads()`):
  - Strings must use double quotes `"`.
  - No extra text outside the JSON object.
  - No trailing commas.
  - Keys and structure must match the required schema.

</instruction>

<output_rules>

- Do not include **explanations, comments, or text** outside the JSON object.
- Ensure the **JSON is syntactically correct**.
- Use only the structure above—**no extra keys or fields**.

</output_rules>

<output_example>

```json
{
  "keyword": "keyword",
  "topic": "topic",
  "language": "language",
  "prompt": "generated text",
  "label": "generated sentiment (which can be positive, negative, neutral)"
}
```

</output_example>
