Release Notes - PoolParty 7.0

As of version 7 PoolParty will not support Internet Explorer anymore. Together with Chrome and Firefox now MS Edge will be instead, see: PoolParty Client Requirements.

As has been defined by the Open Web Application Security Project (OWASP) on their website, web applications should comply to certain standards to make them secure . In this version we hardened PoolParty further, using the following measures:

All those settings are configured via the The poolparty.properties File. Optional settings have to be enabled explicitly so they become effective.

PoolParty version 7 offers the ability to manage the project data and system wide data in all graph databases connected to PoolParty set up in the Semantic Middleware Configurator which support the remote storage functionality.

You can select the remote store in the Advanced tab of the create project dialogue:

In addition in the project's Details View the selected graph database (1) as well as the PoolParty version of the project data (2) are being displayed:

With this new feature all data created in PoolParty can be stored in a separate graph database offering the ability to completely separate the application and the data layer and by that provide more support for high availability and scalability scenarios.

The Ontology Management has seen major improvements in this release. The interface has changed and ontologies as well as custom schemes are being presented in a hierarchical tree now, which adds to the consistency in PoolParty.

Ontologies and custom schemes are grouped into two separate nodes. Below those nodes each ontology or custom scheme has its own node. Expanding them using the Arrow icon beside them lets you look at the child nodes classes, relations and attributes. In the Details View on the ontology or custom scheme level you can see the metadata of the respective ontology or scheme.

Expanding one of those nodes will open the tree of respective classes relations and attributes. Each class, relation or attribute has its own node and Details View.

When you create a new ontology or edit an existing one, a double-click on the respective node will open the Edit Ontology/ New Ontology dialog.

The User Groups tab allows you to assign user groups to your ontology or custom scheme. Based on the user group assignment you can control which users can edit and change the ontology or custom scheme as well as the users that can create data based on that ontology in the Thesaurus Management or the GraphEditor.

The PoolParty Access Management has seen major improvements in this release. The interface users, roles and groups are displayed in a hierarchical tree now, which adds to the consistency in PoolParty.

In addition, the User Management has been redesigned so its consistency has been streamlined once more. When you first access the currently logged in user's node, you will see the hierarchy tree displaying all users, roles and permissions this user can access, in this case a superadmin. On the right find the Details View of the user, with the User Details, Granted Authorizations, Language Settings and Display Settings tabs. In the Granted Authorizations tab you find all roles the user you are looking at has been assigned.

neo4j Integration

As of PoolParty 7 thesauri and ontologies created in PoolParty can be exported to neo4j. You can configure connected neo4j stores using the Semantic Middleware Configurator - External Graph Databases.

The Linking Service for DBpedia has seen considerable improvements in this release:

The SPARQL based DBpedia linking has been replaced to make it more effective. It is now Solr based and uses the spreading activation technique.

The Linked Data Frontend in PoolParty will now resolve HTTP as well as HTTPS requests, regardless of the server's settings. This makes it easier and more comfortable to use because there's no additional administration necessary.

As of this release the concept scoring for corpus analysis has been implemented in PoolParty. This means you will have an additional measure to gauge the validity of concepts found in the corpus. The interface in the Extracted Concepts tab now offers the additional column Relevance (1). It contains the scores of concepts found inside the corpus during the Corpus Analysis. The relevance of concepts here is calculated similarly to the terms in the Extracted Terms List list.

You can use these scores as information about the validity of the extracted concepts in the Extracted Concepts List tab.

The Natural Language Processing ontology (NLP) now included in PoolParty enables additional relations on graph level between concepts, the corpus and the score as well as relevance that has been calculated for them. That way you can use the information later for example for tagging. A remote store configured in the Semantic Middleware Configurator - Overview can be used to export the data and then you would use SPARQL queries to check results and reuse the metadata now available.

Via the Add to ExactMatch button you can now enable exact matching for entity extraction for concepts. When enabled, extraction is only done if one of the labels of the concept exactly matches in the text. For the example below the text "SEMANTiCS" or "i-Semantics" has to match exactly in the text so the extraction is triggered.

Machine learning based Named-entity Recognition (NER) capabilities have been added to the PoolParty Extractor. As of PoolParty 7.0 an organization, people and location extraction has been added that works based on a trained model using a maximum entropy based classification wich in turn is NLP-based to extract organisations, persons and locations. Below a respective API call is shown:

http(s)://<server>/extractor/api/extract?text=Andreas Blumauer is CEO of SWC&nerParamters[0].method=RULE_BASED&nerParameters[0].type=Person

Result:

namedEntities [
        {
                textValues: "Andreas Blumauer",
                frequency: 1,
                type: "person",
]

The Annotate & Annotate and Store services have been greatly enhanced. It allows now to issue directly in the call a SPARQL query to format the extraction result to a specific format needed as output.

In addition the PoolParty Extractor's RDF data model has been extended to cover all the functionality present in the extract call.

The improvements in the APIs include the possibility to define extraction rules. Extraction can now be restricted to concepts that have certain attributes or relations.

The following new DPUs have been implemented: