Vector Services Course

Introduction to the various parts of Vector Services

Ingest Source Coverage and Refresh Schedule#


The following datasets are going away on March 31, 2020


Below are the standard vector sources provided and information on update frequency.

Note: Ingest Sources are subject to rapid change, and both the list of available sources as well as the vector count within each listed source may be different than the given examples. Basic principles still apply.

Index Query Vector Services Overview#

When users know the indexes containing the items they want to query for or retrieve, VectorServices provides the ability to limit operations to specific vector indexes. These calls avoid sending queries to the entire set of vectors, so they tend to take less time to process and return results.

Index Query also supports aggregations. Aggregation capabilities allow users to quickly and easily summarize data based on multiple parameters. Data can be grouped by geohash, date/time, terms, and a number of other aggregation types.

Aggregations are useful for discovering relationships and differences among groups of data. For example, a user could discover the most prolific Twitter users over a set of geohash boxes for the last week by aggregating tweet data by geohash region, time range, and the Twitter user name field. Mixing different aggregation types will give users different views on their data, allowing them to discover relationships they might not have otherwise seen.

See: Vector Services Index Query Reference Overview.

See also:

Write Vector Services Overview#

In order for users to add a vector item or multiple vector items directly to an index, VectorServices provides user vector generation. Vector generation is useful for direct additions of or updates to vectors, without needing a third party service or a vector submission request form. If the user has a few vectors to ingest into VectorServices, using the write API is the preferred method of ingest.

See: Vector Services Write Vectors Reference Overview.

If, however, the user has a large number of vectors to ingest into VectorServices, the bulk ingest process is the preferred method of ingest.

See: Lesson: Bulk Ingesting Vectors into Vector Services

See also: