For certain types of documents, you can associate a static ranking factor that increases the importance of those documents in the search results.
When you create a collection or edit collection settings, you can specify Document importance options. The type of document importance that you select determines whether a static ranking factor is associated with the documents in the collection. When users search the collection that uses static ranking, the static ranking factor influences how documents returned in the search results are ranked.
For Web content, the static ranking factor can be based on links. The number of links to a document from other documents, and the origins of those links, can increase the relevance of that document in the search results.
For documents that include date fields or date metadata, the static ranking factor can be based on the document date. The document date field, which is provided by the crawler, can be the date that the document was last modified or the date that the document was last crawled, depending on how you set up the crawler configuration.
The date of a document might increase its relevance. For example, recent articles in NNTP news groups might be more relevant than older articles. If a data source includes multiple date values, you can choose which one is most important for determining the relevance of documents when you configure the crawler.
For documents that include fields and metadata fields that contain numeric data, the static ranking factor can be based on a parametric field, such as the document size or the date that a document was published.
If you use static ranking with a collection, ensure that you do not mix data sources that use different ranking types in the same collection. For example, if you want to use the links to a document as the static ranking factor, ensure that the collection contains only Web documents. Document ranking is less accurate when sources with different ranking models are combined in the same collection, and the order of the search results might not be as expected.
You should also ensure that documents in the collection contain fields and values that enable static ranking to be applied. For example, imagine a collection that is configured to use static ranking based on document dates, and a crawler in the collection is configured to use a specific field as the document date. If a document does not contain that field, the importance of the document might not be appropriately ranked, and the order of the search results might not be as expected.
Static ranking, along with factors such as assigning a score to boost URI patterns, contributes to the static score of a document and influences the importance of the document. The link-based ranking model is typically applied to Web collections because this model calculates the static rank of a document based on the number of links to the document. A document that is linked to from a high number of other documents is ranked as more relevant.
For this reason, if you configure this model for a non-Web collection or a mixed collection (one that contains Web and non-Web documents), the search quality might be degraded because non-Web documents have no concept of linking.
When you create a collection and select the option to rank by the number of links to documents, link analysis automatically occurs. In link analysis, linked documents are automatically retrieved and indexed with the anchor text in the link. As a result, the linked documents are searchable by keywords that occur in anchor text even if the keywords do not appear in the source document.