Symptom:

When comparing search results with Boolean operators the numbers don't add up.

For example:

• Searching for A returns X records
• Searching for B returns Y records
• Searching for A AND B returns a number of records greater than  X or Y ( the greater number)

Example:

• Searching for 'documentary' returns 21,473 records
• Searching for 'film' returns 477,172 records
• Searching for 'documentary film' returns 24,102 records
• Searching for 'documentary AND film' returns 21,300 records

Explanation:

Primo's search engine has a sophisticated algorithm which expends and alters queries so that the results returned will be relevant to the search.

Such changes relate to searching for:

• authors
• synonyms
• common misspellings
• Phrase searching
• Searching in full text of PCI records
• Etc.

The search engine algorithm builds a series of Boolean searches which increases the relevance of the results, as opposed to an SQL search, for instance, which works in a direct manner on the data.

This is important in particular for searches in PCI where, due to its sheer size, a direct search (as SQL) would return an extremely large amount of records, all meeting the search query but  many of them may have low relevance to the end user.

Conclusion:

The results of different searches aren't necessarily comparable in a linear relation.

• Article last edited: 18-Jan-2017