Currently, deepset Cloud supports the batch question answering job that you can use to process queries in bulk. You can run the query set once on all your files or repeat the queries per each individual file. Once you have the results, you can easily format them to your specifications and share them with anyone you wish without the need to create an account or log in. Just generate a link to an HTML page that you can share.

On the Jobs page, you can view all the jobs created in your workspace. Clicking a job opens its details page, where you can check:

• The pipeline it uses
• The query set
• The number of queries run
• Results

Query Batch

To run a batch question answering job, you must create a CSV file with your queries that you then upload to deepset Cloud when creating a job. The CSV query set contains the queries and any labels or groupings you want to use for your queries on the shareable HTML page. For example, when running a job on property reports, for a query "What is the address of the building?" you may want the label to be "Address". You may also want to group queries like "What is the address of the building" and "How many floors does this building have" under a group name "Property Details".

You can add filters to the query set to narrow down the set of files the job runs on. Filters are simply file metadata. For example, your files can have a metadata key called "type" with values "property report", "news", "reviews," and so on. You may then set a filter {"type": "property report"} to run the job on property reports only.

Filters also make it possible to run the same query against multiple files. Simply repeat the query in your query set specifying the file name in the filters column. For examples of query sets and how they impact the results page, see Prepare a Query Set.

For more information on filters, see Filtering Logic.