Bookshelf Home | Contents | Index | PDF

### Relevance Method Setting

This setting determines the retrieval model and relevance algorithm to be used in search calculations. Document relevance is used to describe how closely a given document matches a query. The relevance of a document can determine the order of the search result that is displayed.

There are three retrieval models that can be specified. Each model has a different impact on the search results when used with a ranking algorithm. The RELEVANCE function combines the retrieval model and ranking algorithms to provide a calculated relevance value that estimates the similarity between the row and the query criteria.

##### Example

Property: Relevance Method

Value: V2:4

There are three retrieval models that can be specified:

• Strict Boolean. When searching for matching rows in a table, this model uses strict Boolean operations for combining search terms. There is no measurement of the degree to which a match is found.
• Fuzzy Boolean (F). Similar to the strict Boolean model, this model differs in that it does not rule out rows that contain only some of the terms.
• Vector Space (V). This retrieval model relaxes the strict Boolean model by allowing the match to be measured by using some of the search terms. The relevance value is based on the proportion of the most relevant search terms that are matched in the row.

Table 34 lists the options available when setting up relevance ranking.

Table 34. Relevance Ranking Options
Values
Relevance Ranking
Description

2:1

Hits Count (Algorithm 1)

Counts the total number of occurrences of the individual words (not phrases) matched, regardless of the term frequency within the table. This value is only valid when the retrieval method is set to Boolean.

2:2

Terms Count (Algorithm 2)

Counts the number of search terms matched. The frequency of the occurrences of the terms is not considered. This value is only valid when the retrieval method is set to Boolean.

2:3

Terms Ordered (Algorithm 3)

Uses a mathematical formula that computes the relevance statistically. It combines the characteristics of algorithms 1 and 2, and takes into account not only the number of occurrences of each search term, but also a statistical measurement of how common the term is over all the rows in the table (document frequency). This value is valid using the following retrieval methods: Boolean, Fuzzy, or Vector. The optimized retrieval methods are Fuzzy or Vector.

2:4

Critical Terms Ordered (Algorithm 4)

Uses a mathematical formula that computes the relevance statistically, and accentuates the effect of the inverse document frequency. It squares the search term importance before multiplying it with the number of occurrences of the search term. This value is valid using the following retrieval methods: Boolean, Fuzzy, or Vector. The optimized retrieval methods are Fuzzy or Vector.

To administer the relevance method

1. Navigate to Administration - Search > Advanced Settings.
2. From the Setting Properties menu, select New Record.

A new row appears in the Setting Properties list.

3. Click in the Name column of the new row, and from the drop-down list, select Relevance Method.
4. Click in the Value column of the new row, and from the drop-down list, select the appropriate relevance setting, then step off the record to save it.