Skip to content

Latest commit

 

History

History
163 lines (111 loc) · 8.24 KB

setup.md

File metadata and controls

163 lines (111 loc) · 8.24 KB
Raw

TermIt Setup Guide

This guide provides information on how to build and deploy TermIt.

Build

System Requirements

  • JDK 8 or newer (tested up to JDK 11 LTS)
  • Apache Maven 3.5.x or newer

Setup

Maven Profiles

To build TermIt for non-development deployment, use Maven and select the production profile.

In addition, full text search in TermIt supports three modes:

  1. Default label-based substring matching
  2. RDF4J repository with Lucene index
  3. GraphDB repository with Lucene index

Options 2. and 3. have their respective Maven profiles - rdf4j and graphdb. Select one of them or let the system use the default one.

Moreover, TermIt can be packaged either as an executable JAR (using Spring Boot) or as a WAR that can be deployed in any Servlet API 4-compatible application server. Maven profiles standalone (active by default) and war can be used to activate them respectively.

Application Configuration

The rest of the configuration is done in the application.yml file in src/main/resources.

Most of the parameters there should be self-explanatory or have documentation in the ConfigParam class.

There is one parameter not used by the application itself, but by Spring - spring.profiles.active. There are several Spring profiles currently used by the application:

  • lucene - decides whether Lucene text indexing is enabled and should be used in full text search queries.
  • admin-registration-only - decides whether new users can be registered only by application admin, or whether anyone can register.
  • no-cache - disables EhCache which is used to cache lists of resources and vocabularies for faster retrieval.

The lucene Spring profile is activated automatically by the rdf4j and graphdb Maven profiles. admin-registration-only and no-cache have to be added either in application.yml directly, or one can pass the parameter to Maven build, e.g.:

  • mvn clean package -P graphdb "-Dspring.profiles.active=lucene,admin-registration-only"

Example

  • mvn clean package -B -P production,graphdb "-Ddeployment=DEV"
  • clean package -B -P production,rdf4j,war "-Ddeployment=STAGE"

The deployment parameter is used to parameterize log messages and JMX beans and is important in case multiple deployments of TermIt are running in the same Tomcat.

Building on Windows

Building TermIt on Windows sometimes requires Maven to be configured to use the UTF-8 encoding. In certain cases, Maven would otherwise use the default Windows encoding (e.g., CP-1250) which can cause issues when accessing URLs with Czech accents (e.g. the ontology/termit-glosář.ttl file required for generating application vocabulary).

This configuration can be done, for example, via the MAVEN_OPTS environmental variable. Either set it temporarily by calling

  • set MAVEN_OPTS= -Dfile.encoding="UTF-8"

or configure it permanently by setting the MAVEN_OPTS variable in System Settings.

Deployment (WAR)

System Requirements

  • JDK or later (tested with JDK 11)
  • Apache Tomcat 8.5 or later (9.x is recommended) or any Servlet API 4-compatible application server

Setup

Application deployment is simple - just deploy the WAR file created by Maven to the application server.

What is important is the correct setup of the repository. We will describe two options:

  1. GraphDB
  2. RDF4J

GraphDB

In order to support inference used by the application, a custom ruleset has to be specified for the TermIt repository.

  1. Start by creating a GraphDB repository with custom ruleset
  2. Use the ruleset provided in TermIt at rulesets/rules-termit-graphdb.pie
  3. Create the repository, configure it as you like (ensure the repository ID matches TermIt repository configuration)
  4. Create the following Lucene connectors in GraphDB:

Language can be set for each connector. This is useful in case the data contain labels, definitions, and comments in multiple languages. In this case, if connector language is not specified, FTS results will contain text snippets with text in all the available languages. For instance, assume there is a term with label území@cs and area@en. Now, if no language is specified for the label connector, the resulting text snippet may look as follows: <em>území</em> area, which may not be desired. If the connector language is set to cs, the result snippet will contain only <em>území</em>. See the documentation for more details.

RDF4J

In order to support the inference used by the application, new rules need to be added to RDF4J because its own RDFS rule engine does not support OWL stuff like inverse properties (which are used in the model).

For RDF4J 2.x:

  1. Start by creating an RDF4J repository of type RDFS+SPIN with Lucene support
  2. Upload SPIN rules from rulesets/rules-termit-spin.ttl into the repository
  3. There is no need to configure Lucene connectors, it by default indexes all properties in RDF4J (alternatively, it is possible to upload a repository configuration directly into the system repository - see examples at [1]

For RDF4J 3.x:

  1. Start by creating an RDF4J repository with RDFS and SPIN inference and Lucene support
    • Copy repository configuration into the appropriate directory, as described at [2]
    • Native store with RDFS+SPIN and Lucene sample configuration is at [3]
  2. Upload SPIN rules from rulesets/rules-termit-spin.ttl into the repository
  3. There is no need to configure Lucene connectors, it by default indexes all properties in RDF4J

Common

TermIt needs the repository to provide some inference. Beside loading the appropriate rulesets (see above), it is also necessary to load the ontological models into the repository.

  1. Upload the following RDF files into the newly created repository:

    • ontology/termit-glosář.ttl
    • ontology/termit-model.ttl
    • http://onto.fel.cvut.cz/ontologies/slovník/agendový/popis-dat/model
    • http://onto.fel.cvut.cz/ontologies/slovník/agendový/popis-dat/glosář
    • https://www.w3.org/TR/skos-reference/skos.rdf

  2. Deploy TermIt. It will generate a default admin account, write its credentials to standard output and into a hidden file in the current user's home.

For proper functionality of the text analysis service, Annotace has to be running and its URL configured in application.yml.

Repository Credentials

Note that if the repository server to which TermIt is connecting is secured (strongly recommended), it is necessary to put the repository user username and password into application.yml so that TermIt persistence can access the repository. The configuration parameter are in Configuration.Repository class, but here's a quick example:

termit:
    repository:
        url: http://localhost:7200/repositories/termit
        username: termit
        password: supersecretpassword