Use the new DISCO sampling feature is to preview a randomized sample of documents that you plan to produce.
Sampling Search Syntax
- sample(0.1, “any query”) - gives you 10% of search results set
- sample(500, “any query”) - gives you a maximum of 500 documents from search results set
- sample(10%, “any query”) - gives you 10% of search results set
For example:
sample(500, tag(Responsive) % tag(Attorney-Client) % tag(Work-Product))
In plain English, this search would find a maximum size 500 document sampling of all documents tagged Responsive but not tagged Attorney-Client or Work-Product. Such a sample could be used for quality control to check the contents of the production set qualitatively. Upcoming releases of the software will also include tools to assess the contents of the production set quantitatively by computing metrics such as prevalence, precision, and recall.
From your search results, tag the documents.
Some important items to note:
- IF you search or refresh the page, it will give you a new random subset of documents.
- IF you give it a number that is more than the search results, it will give you the total number of search results.
- IF you use the sample query, sort order for search results is disabled.
Comments