Below are some key terms used across our services and user documentation.
Projects are top-level containers that can contain multiple search collections. Projects store information about billing, authorized users, usage statistics and Project level API credentials.
Collections contain the data that you want to search through. Collections contain the configuration associated with your data including rules, synonyms, and authorized domains as well as collection-level analytics. Each collection has an associated schema that designates field names, field types, and whether a field's data is indexed for text search.
The console is the logged-in area of the Sajari website where you can manage your search project and collections.
A pipeline is a grouping of pre-defined data processing steps performed to run a search query.
A record is a single entry in a collection, made up of structured key-value pairs. For example:
'firstname' (key) = 'John Doe' (value), or
'publishedDate' = '2020-01-02'.
Records are analogous to a row in a database. Records can represent:
- Standard documents (e.g. .HTML, .PDF, .DOC)
- Custom data objects (if your data is not webpages or documents)
The data structure of Record is called a Schema. Webpages and documents have a pre-defined Schema. For custom Records, you can define your own Schema.
A field is a sub-component of a record. Fields are analogous to a column in a database.
The schema is a blueprint of fields, properties and constraints of your records.
Click-through rate (CTR) is a metric used to measure search success. CTR is reflected as a percentage, made up of the total number of queries divided by the number of result clicks. For example, if the query "ice cream" is searched for 20 times and five of those searches result in a click, the CTR is 25%.
Operations reflect any time a record in a collection is added, edited, or deleted. Search queries to a collection are also counted as an operation. We do not calculate price using operations.
A search query is a piece of information used as an input to find a set of records that are most relevant to it.
Filters are broad classifiers used to exclude search results. Filters are most often used in a search interface to remove unwanted search results, and remain consistent for all queries. Filters can also be used to restrict what search results are returned from a specific interface.
Similar to filters, facets are classifiers used to exclude results that do not match certain criteria. Facets are most often used when search results are returned, and allow the end-user to select what results should are displayed. Unlike filters, facets are generated from the characteristics of the returned search results and change depending on the query.