You can import data from the complete undeleted history of Twitter via Sifter.
Create a query and get an estimate of the data retrieval cost. After you receive the estimate, you can proceed with the transaction or refine the query and then resubmit the estimate. You can submit an unlimited number of free estimates.
- Create a project, or open an existing one from the Project list in the Navigation Sidebar.
- In the Project Options section, click Import Data.
- In the Twitter section, click Historical via Sifter.
You are redirected to the Sifter login page.
- Log in to Sifter.
Note: If you do not have an account yet, click Register at the bottom of the page.
- Click New estimate.
- In the Job Title field, type a name for the new estimate, and then click Continue.
- Enter the start and end dates. A price estimate based on the number of days is displayed. Note: You must enter a valid date range. The most affordable estimates cover days or weeks not years.
- To create the query that you want to use, do one of the following:
Option Steps To manually create the filter
- In the Rule Text field, type the query.
To create a Gnip PowerTrack filter in the interactive filter
- From the All/Any list, select an option.
- From the Rule list, select the rule, and then enter a value in the blank field.
- To add another rule, click the Add button.
- To add the filter to the Rule Text field, click Translate.
For more information on creating queries, watch the Using Gnip PowerTrack Filters in Sifter video.
- When you are satified with the query, click Complete.
- In the confirmation dialog, click Submit.
When the estimate is ready, you receive an email notification, which contains the approximate number of tweets that fit the query and the cost of the data retrieval. If you accept the estimate and purchase the license to use the data, it is saved in your DiscoverText account.
Note: If the estimate contains a yellow warning message, there may not be any data for that query.
Note: The actual number of data items (activities) that are retrieved may be (and usually are) different than the estimate. See this FAQ for more information.