Twitter Sentiment Annotation Help
Budget: $15 – $25 AUD
A collection of public tweets now sits in a CSV waiting for a clear, binary sentiment label. Each row contains the tweet text, ID and time stamp; the goal is to assign “positive” or “negative” to every entry so the file can drive a supervised-learning model later on.
I will share detailed labeling guidelines, edge-case examples and a small calibration batch to make sure we interpret sarcasm, emojis and slang the same way. The work can be done in Excel, Google Sheets or any lightweight annotation tool you prefer—as long as the final file comes back in the same structure with an added sentiment column.
Deliverables
• Fully annotated CSV of all supplied tweets, one sentiment tag per row
• A short note summarising ambiguous cases or guideline gaps you encountered
Acceptance criteria
• ≥ 95 % agreement with the gold-standard sample I will spot-check
• No rows added, removed or reordered; only the extra sentiment column
Clean, consistent, on-time annotations are all that’s needed—no model training or extra analysis at this stage.
I will share detailed labeling guidelines, edge-case examples and a small calibration batch to make sure we interpret sarcasm, emojis and slang the same way. The work can be done in Excel, Google Sheets or any lightweight annotation tool you prefer—as long as the final file comes back in the same structure with an added sentiment column.
Deliverables
• Fully annotated CSV of all supplied tweets, one sentiment tag per row
• A short note summarising ambiguous cases or guideline gaps you encountered
Acceptance criteria
• ≥ 95 % agreement with the gold-standard sample I will spot-check
• No rows added, removed or reordered; only the extra sentiment column
Clean, consistent, on-time annotations are all that’s needed—no model training or extra analysis at this stage.