How to use Semantic Search?

The Jobscentral database runs Keyword and Semantic search. We advise to use the word clouds to enter filter criteria!

Boolean Search vs Semantic Search

Boolean: Boolean queries are constructed by using keywords combined with AND/OR/NOT operators. Statements are often nested to include all the necessary keywords and can get complex quickly. When using multiple keywords they all have the same relative importance. Only candidates that are an exact match for the Boolean statement will be returned.

Semantic: Semantic search is able to utilize functionality beyond simple keywords and operators to return ranked results that are similar to if a person had compiled them. When performing a semantic search all search terms are provided and then ranked or weighted to identify how relevant they are for the search. This permits candidates in the search results to be ranked based on how closely they match the criteria.

Keyword Relevance

When performing semantic searches it is possible to indicate the relevance of each keyword in the search query. Marking certain keywords as more relevant than others influences the ranking of the results that are returned. There are three possible levels of relevance: Must Have, Should Have, and Nice to Have.

Must Have: Use Must Have to indicate that the term is required to be present in all search results. Any candidates that do not include the Must Have item will not be returned in the results.
1. Set by default.
Should Have: Most search keywords will be marked as Should Have. Candidates that include the Should Have item will be ranked higher in the results than candidates that do not have it. Should Have allows results that do not include the term to be returned but they will be ranked lower.
1. Typing “^” in front of a keyword gives the option to rank the result higher if a certain keyword is added after “^”. (The should-have query parts are marked with condition STRONGLY_FAVORED.)
  Example:
  - Engineer ^manager
Nice to Have: Use Nice to Have to boost the ranking of items that are not mandatory but would make one candidate a better fit over another. The Nice to Have ranking weight is less than Should Have but still higher than candidates that do not have the term.
1. Typing “%” in front of a keyword gives the option to rank the result higher if a certain keyword is added after “%”. (The nice-to-have query parts are marked with condition FAVORED.)
  Example:
  - Engineer %manager

Typing "AND" or leaving a blank space (spacebar) or a comma (,) will add an "&" command to your search

Examples:

Engineer AND html
Engineer html
Engineer,html

Typing “OR” or put your keywords in brackets “(keyword1 keyword2)” will add a "/" command to your search

Example:

Engineer OR manager
(Engineer manager)

Typing "NOT" or adding a minus symbol (-) will add a "NOT" command to your search

Example:

Engineer - manager
Engineer NOT manager

Now let’s mix them up!

Example: Engineering OR Engineer AND manager NOT team

Further information:

Keywords- E.g. Java developer Amsterdam. Keywords are not case sensitive, so "Java", "JAVA", and "java" are the same.

Phrases- Phrases enclosed by quotes are used to match a sequence of words. Special characters are ignored inside a phrase.

Proximity- Query terms within [ ] match if they occur in the document in any order, possibly with one or two words in between. This is called proximity matching. Proximity matching is more flexible than phrases. For example, the above query [Java developer] also matches documents containing "java software developer" and "java enterprise software developer". Proximity queries can only contain simple keywords, but not phrases. Quoting inside proximity queries will be ignored.

Boolean Search- The query language has a limited support for commonly used Boolean expressions. The above examples are interpreted as expected, respectively searching for a document mentioning the keywords “developer” and either “java” or “.net”, and a document mentioning the keywords “Perl” and “C++” but not the keyword “Java”.

Boolean operators are only interpreted in simple full-text queries. If the query contains any other above mentioned query syntax, such as field names or condition operators, boolean operators won't be recognized.

Wildcards- Search + Match allows wildcard queries. If a query term ends with a trailing * it gets expanded to the most common completions. Documents containing any of the found completions will match. For example, the above query develop* will match on developer, development, developed, etc. A few restrictions apply:

The * symbol can only be used at the end of a query term.
The * symbol needs to be preceded by at least 2 characters.
Wildcard terms cannot be part of phrases or proximity expressions.

When using the filters:

Use the + or – button on any “word cloud” to include (+) or exclude (-) skills, positions, and other filter options