Search Service Assembly

The service can be assembled in various ways for the end user of the solution, depending on how it was designed by the portal administrator.

To use the Search service, simply instantiate it in a channel and drag its interfaces to the areas of the pages in edit mode. Its interfaces correspond to areas of interaction with the service by the end user.

The service offers two interfaces aimed at facilitating the assembly of a solution. These interfaces allow the search engine to be used on any portal page, with results displayed through a list of contents, allowing users to refine their search from the results screen.

Search: This interface allows users to type the content to be searched;
Search with Results: This interface allows users to view the search results, which are titles of contents linkable to their details. If the search result is a link to a document, the details page of this document will be displayed instead of triggering its download. Additionally, this interface allows the user to perform a new search.

To make the Search service available to the end user, the searchable services (Announcements, News, Documents, Images...) must be instantiated in the same channel that contains the search service instance, or in channels below it. The search interfaces can be instantiated on the same page or on different pages.

As an example, consider the following channel hierarchy containing instances of Search and News services:

In this example, a search performed by the Search A instance will find contents in the News instances M, N, O, P, Q, R, and S. The Search B instance will only search in News instances M and N. The Search C instance will find contents from instances P, Q, and R. Finally, the Search D instance will only search for contents in News instance S.

Parameterization

It is possible to specify whether the "Search Result" will display the instance path, the instance name, and/or the service instance image along with the returned content.

To do this, it is necessary to configure in the context menu of the interface (accessed by right-clicking) in the Fields option, respectively for the fields Service Path, Service Name, and Service Image. From there, the default XSL of the service will begin to read these available information in the XML.

The figure above exemplifies which information is being passed to the XML, however, the default XSL of the search service is implemented to return only the content title (Title), content introduction (Subtitle), Service Name, and Service Path. Although Address (URL) and Relevance are marked by default, they are not being informed in the XML and therefore the XSL is not displaying this information on the interface.

An important observation is that these last two information when added to the search result increase processing since for each returned content, these informations will be obtained from the cache or from the database if there is no cache.

Another important observation is that it is recommended to use the information of Service Name and Service Path if the solution assembly architecture provides suggestive names. This means it is not advisable to use names for service instances such as "News 1", "News 2". Additionally, the service instances cannot be in one channel and their interfaces in pages of another channel.

Search Result Configurations

In the context menu of the Search Result interface, there is the option “Search Result Configurations”. This option allows the grouping properties to be customized. By default, service groupings, service instance, categorization, and tags will be active.

The grouping will occur in the Big Data repository.

Sorting

The Search Result interface provides in its context menu the option "Sorting". Through it, it is possible to configure a sorting for the "Search Result" interface. By default, the interface sorts the results by the "Relevance" of the content.

The configuration options available are:

Relevance
Title A-Z
Title Z-A
Most Recent
Oldest

Service Instance Configurations:

Field weight modifiers for search:

In the service instance, it is possible to configure (through its menu, in the Portal Studio) field weight modifiers for the search.

By default, the following weights are applied:

lum_standard.title (Content Titles): 10
lum_standard.categories.fullCategories.name (Content Categories): 7
lum_standard.tags.analyzedName (Content Tags): 7
lum_standard.introduction (Introduction): 5
Other fields: 1

The weight applied to a field must be positive. Values between 0 and 1 decrease relevance in the search. Values above 1 increase relevance in the search.

The customized values will be unified with the default values (giving preference to the customized ones) to generate the final values to be used in the search. That is, if the following values exist in the registration:

lum_standard.title: 2
my.bigdata.field: 4

The values used at the time of the search will be:

lum_standard.title: 2
my.bigdata.field: 4
lum_standard.categories.fullCategories.name: 7
lum_standard.tags.analyzedName: 7
lum_standard.introduction: 5

Semantic search:

In the service instance, it is possible to configure (through its menu, in the Portal Studio) the behavior of the semantic search of LumisXP.

By default, the following values are applied:

Number of results obtained in the semantic search: 10
Weight of semantics in calculating relevance: 0.5
Minimum similarity: 0.78

For more information on how each of these parameters will be used, refer to the section Semantic search.

Notes

For information on how to instantiate channels and pages, please refer to the section Portal Studio;
Not all fields (title, description, introduction, date...) of content are searchable (returned by the search engine). This needs to be explicitly stated in the service definition. For content management services to be created by the creation wizard, for example, it is possible to determine which fields will be searchable. For more information consult the section on creating services using the Content Assistant;
The search returns the closest destination, which means that if a details interface of the same service instance of a searchable content has been instantiated on different pages, the link returned will point to the page that is closest to the channel containing the search service instance;
Check the section on Reindexing Content so that the search engine works correctly;
Through the right-click of the interface, other available parameterizations for the service instance and customizations for the interface can be found. Refer to the section Customization of Interface Instances.