Manage rank profiles (FAST Search Server 2010 for SharePoint)

 

Applies to: FAST Search Server 2010

The rank profile is configured to calculate the rank score with input from quality rank components (also known as static rank components) and dynamic rank components. The default rank profile is named default, which is also used as the default sort criteria (Sort by Relevance) on the search result page. Be aware that you tune rank components independently of one another. The value of two rank components with the same weight is not necessarily the same so that changing a weight from 50 to 100 on one component could lead to more effect on the final rank score than changing a weight from 50 to 100 on another component.

The rank profile

Optimize search relevance in FAST Search Server

Rank Factor Description

Quality (static) rank components

Quality

Quality rank points, also known as static rank points, are derived from the following managed properties:

  • Urldepthrank: Rank points given to boost shorter URLs.

  • Docrank: Rank points given based on the number of and relative importance of links pointing to an item.

  • Siterank: Rank points given based on the number of and relative importance of links pointing to the items on a site.

  • Hwboost: FAST Search Server 2010 for SharePoint placeholder for generic usage of static / quality rank points.

In addition to these, you can add any custom integer managed property to the list of static/quality rank components.

Dynamic rank components

Freshness

Rank points given based on the age (last modified) of the item compared to the time of the search. The source for the last modified time is the managed property named "Write".

The resolutions used (second/minutes/hour/day/year) are important for the ranking, because a lower resolution implies that smaller time differences will affect the ranking.

Context

Rank points given based on where in the item the search words hit. Context boost is assigned as an overall weight and as a weight for an individual managed property. The different managed properties such as title and body are mapped into different importance levels. This makes it possible to give more rank points when the search word hit in a title instead of in the body text of the item.

Proximity

Rank points given depending on the distance between the search words and where these words are located in the search item. For example, if two search words appear in the same managed property within the full-text index, they will get more rank points.

Managed property field boost

Rank point given based on a specific item type of a managed property. For example, if you want to give additional rank points to indexed items of a specific type such as Word documents, you would give search items with the file extension .docx additional rank points.

Authority (anchor text)

Rank points given based on when a search word retrieves hits in the link text (anchor text). This rank boost has two components:

  • Partial match - the link text is Microsoft Word, and the search is Microsoft.

  • Complete match - additional score if there is a match between the search word and link text.

Query authority (click-through)

Rank points given based on when a search retrieves hits in items associated with previously clicked on search results (click-through). In the search front-end, all of a user’s searches and the items that the user clicks on that are associated with those searches are stored and processed. For example, if a user searched for vacation and clicked on the vacation request template in the search result (e.g., http://myserver/template/vacation.aspx), this URL is associated with the search word and receives additional rank points when any new search with the same search word is performed.

Processing occurs one time each day. The last 30 days of clicks are kept for processing.

This rank score also has two components:

  • Partial match - the initial search words that resulted in a click to this item was Microsoft Word, and the search word now is Microsoft.

  • Complete match - an additional score if there is a match between the search word and link text.

In this section:

See also

The index schema and its features (FAST Search Server 2010 for SharePoint)