Oracle Endeca Commerce Relevance Ranking

Relevance ranking refers to ordering search results by criteria that are relative to the search term. This is in contrast to common use of the terms “sorting” and “static sorting”. This is a recursive exercise to analyze and implement appropriate Relevance Ranking for an application for getting desired results. There are different Relevance Ranking module available for ordering search results which are described below.

Exact: Exact module provides a finer grained, but more computationally expensive alternative to the Phrase module. It groups results into three strata:

1. Results whose complete text matches search terms exactly.

2. Results that contain search terms as an exact substring.

3. Other hits, such as normal conjunctive matches. Any match requiring query expansion lands in this lowest stratum

Field: Field module ranks records based on field priority defined for search interface in which it matches. Only the best field in which a match occurs is considered.

First: First module ranks records based on how close a match is to the beginning of a field. It is primarily for unstructured data.

Freq: Frequency (Freq) module provides result scoring based on the frequency (number of occurrences) of the search terms within a record. Results with more occurrences of the user search terms are considered more relevant.

Glom: Glom module ranks single-field matches ahead of cross-field matches. It serves as a useful tie-breaker in combination with the Maximum Field (Maxfield) module

Interp: Interpreted (Interp) module scores records based on processing techniques used to obtain match. Techniques considered include partial matching, cross-field matching, spelling correction, thesaurus and stemming.

Maxfield: Maximum Field (Maxfield) module behaves identically to the Field module, except in how it scores cross-field matches. Unlike Field, which assigns a static score to cross-field matches, Maximum Field selects score of the highest-ranked field that contributes to the match.

Nterms: The Number of Terms (Nterms) module ranks matches according to how many search terms they match. For example, in a three-word query, results that match all three words will be ranked above results that match only two, which will be ranked above results that match only one.

Numfields: Number of Fields (Numfields) module ranks records based on number of complete matching fields defined in the applied search interface. Only whole-field, rather than cross-field, matches count. Therefore, a result that matches two fields matches each field completely, while a cross-field match typically does not match any field completely.

Phrase: Phrase module considers search terms as an exact phrase, or a subset of an exact phrase, more relevant than matches simply containing search terms scattered throughout the text. Phrase module takes the following options:

Rank by subphrase length

Approximate subphrase matching

Apply spelling, thesaurus and stemming

Consider field ranks

Proximity: Proximity module ranks based on how close search terms in a multi-term query appear to each other in a field. It is primarily for unstructured data.

Spell: Spell module ranks true matches ahead of spelling-corrected matches.

Static: Static module ranks records by an attribute in either ascending or descending order. You have to define an attribute which can be set for ascending or descending sorting.

Stem: Stem module ranks true matches ahead of matches due to stemming.

Stratify: Stratify module is used to boost or bury records in the result set into various strata using the Endeca Query Language (EQL). Example: collection()/record[P_Rating>3],*,collection()/record[P_Price<50]

Thesaurus: The Thesaurus module ranks true matches ahead of matches due to thesaurus entries.

Wfreq: Like Frequency (Freq) module, Weighted Frequency (Wfreq) module scores results based on the frequency of search terms within a record. Additionally, the Weighted Frequency module weights the terms for each result by the overall frequency in the complete data set. Less frequent terms (that is, terms that would result in fewer search results) are weighted more heavily than more frequently occurring terms.