Summary: The search query matches specific information in a document, however that document is returned far down in search results, how can I make this document more relevant for my specific search query.
Cause: The Google Search Appliance uses internal proprietary algorithms to determine how documents get ranked. The process is done at both crawl time and serve time, it's possible that either the initial page rank associated with the documents is not strong enough or the search query is not being considered relevant enough to return the documents first.
Troubleshooting Steps: As described in the official documentation's xml reference the SCOREBIAS tag can be used to view PatternScorer, DateBias or MetaData score of the search results. In the same document you can also use the entsp parameter to view the advanced relevance scoring parameters as set in the admin console.
Workaround: A possible work around to the issue is to use the key match feature, which will ensure that particular documents get prioritized when the search keyword is matched in the document
Fix: If this is a systemic issue in your environment it may be better to create collections with specific contents more suitable for users performing these queries. As such you can limit the number or relevant documents that can be displayed eliminating irrelevant documents.
It is also possible to alter the bias of documents which will skew the results in favor of the bias setting. There are three mechanisms to achieve this as described below.
- Source biasing--enables you to affect scores of documents that belong to a specified collection or that match certain URL patterns
- Date biasing--enables you to affect scores of documents by date
- Metadata biasing--enables you to affect scores of documents based on metadata associated with documents
Additional Information: Using Result Biasing to Influence Result Ranking
Known Issues: 6081241 - the ScoreBias value for DateBias always returns VALUE="-1".
Versions affected: All.
Versions fixed: Unfixed.