PoolParty GraphSearch provides an application layer on top of the knowledge graph you created and your enriched content, once you have applied the knowledge graph to it. It offers standard functionalities, like search suggestions and facets, but also more advanced functionalities like recommendations and content suggestions (similarity) and both of them can be adapted to specific use cases.
PoolParty GraphSearch offers two main approaches for an application based search: taxonomy based and ontology based.
Taxonomy based search can be seen as the classical workflow, where textual content is tagged and stored in a high performance search index like Elasticsearch or Solr. The tags are based on a PoolParty taxonomy. This approach fits best when a document based data model is available: unstructured or semi-structured data that can be made searchable and visualized, based on a controlled and consistent vocabulary. Optional custom fields allow to introduce additional search options and allow to modify according to specific requirements of the search application.
In this approach, the taxonomy is the backbone for an easy to use faceted search, supporting a drill down search. The search experience for users improves drastically, compared to classical full text search solutions due to the usage of synonyms and the providing of contextual information, which in turn are based on the hierarchical structure of the taxonomy during search time.
Ontology based search allows to build search applications, based on knowledge graphs which are stored in triple stores. Due to the nature of knowledge graphs, the model to define search objects allows a lot of freedom. In PoolParty GraphSearch this freedom of data modelling can be used to cover nearly unlimited search scenarios. The advantage is that the modelling needs to be defined only once in PoolParty Ontology Management and can be applied simply by way of the configuration during setup of a search application. Since in this approach the RDF data model is applied, a data integration of large publicly available data sets like Wikidata, DBpedia, Geonames etc. can be provided in a dynamic way.
Faceted search makes use of classes, relations and attributes that are defined in a PoolParty ontology. It can be used by the user to retrieve the requested information through drill down. For a data model, where different business object types (person, company, product, location, and others) are central, the ontology based approach fits best. Therefore, this approach is best to build search applications and visualisations where highly structured data is available.
Custom plugins can be implemented for recommendation and similarity search. These plugins can be designed to improve the user search experience by data object traversal and providing search results that fit the context.
Since all relevant data can be acquired from PoolParty and graph databases, interactive visualizations or other forms of analytics, such as reports, can be built using SPARQL queries. Customizing integrated linked knowledge graphs and adapting SPARQL queries allows you to adapt and modify your analysis applications in a very dynamic and agile way. The same way you do mashups, you can combine different data sets and formulate queries to retrieve data according to your needs.