Generate Labeled NLP Data

We are generating an NLP dataset involving sentence pairs labeled to be contradictory or not contradictory. We would like to hire a freelancer who is highly fluent in English and able to write and label pairs with different qualities, such as opposites (The pan is hot / The pan is cool), changed order (The first step is to set the lever to position one / Setting the lever to position one is the final step), and consistent paraphrasing (The writer of the document should be the one to upload it to the server. / Upload to the server must be done by the document author.). We will provide a generated set of sample sentences that you will modify to meet the given criteria and label based on a provided key. The modified pairs and labels should be stored in csv format. All of the sentences will be of a technical style, so precision is important. Online paraphrasing tools such as [login to view URL] may be used to speed up your process, so long as the appropriate modifications are made and you read each pair to verify the correct labels are applied given the context as the paraphrasing tools are not perfect. For this first task, we would like to commission 2,000 sentence pairs, with hopes to establish a good relationship for further work. The generation of each pair should take 20-30 seconds for a fluent speaker for a total time of around 16 hours.