Semantic Web Ontology Portal Evaluation Approach

From Earth Science Information Partners (ESIP)

Introduction

This page is no longer the active planning document for evaluating possible ESIP ontoogy repository instances. The page you want for that is https://esipfed.github.io/stc/UseCases/STCUseCasesAndRequirements.html.

This wiki page supports ESIP's evaluation of semantic repository solutions in particular providing insight into;

  1. what current and future ESIP Semantic Technologies Committee (STC) community requirements (CR) are,
  2. a comparative evaluation of different technology stacks/approaches; with the aim of assessing different solutions based upon how well they satisfy CR as defined within this document, and
  3. a business cost justification model that could be used to justify funding semantic web technology stacks.

All ESIP members (and anyone else with an account on this wiki) are encouraged to contribute towards this document.

Semantic Repository Implementation Requirements

In general in this document:

  • 'ontology' refers to any semantic artifact, whether expressed in OWL/RDF, another OWL representation, or a simpler form like Excel or CSV that can be converted into a semantic artifact.
  • URI is used to mean a unique web identifier, aka IRI; it usually but not always is a URL.

Features/Services

These are the functions we'd like to see offered by the repository:

Ingest capabilities

  • It should be possible for a person with an ESIP account to upload their semantic resources to the repository
  • It should be possible for a person with an ESIP account to upload new versions of their semantic resources to the repository.

Access capabilities

  • Publicly accessible resources in the should be accessible via a Permanent Identifier (PID) (such as a DOI or equivalent) at both the resource and individual concept level which implies namespaces, versioning, and redirect capabilities
  • When a Permanent Identifier technology dies (e.g., PURL's from OCLC) mechanisms to "fail over" to an alternate PID scheme relatively easily should be supported. This might be accomplished by supporting multiple PID technologies simultaneously and allowing resources to have multiple PID's at a variety of levels.

Search capabilities

User-specific search and update capabilities

  • A logged in user (i.e., one with an ESIP account) should be able to query to obtain a list of their own resources in the repository
  • A logged in user needs to be able to update their own information in the repository (i.e., profile including things like passwords, email addresses, etc.). Relevant updates need to proliferate to the resources that user has uploaded.

General search and update capabilities

Browse capabilities

Display capabilities

  • Display a list of semantic resources (e.g., ontologies)
  • Display the terms and relationships for a particular version of a semantic resource

Download capabilities

Mapping capabilities

Inferencing capabilities

Query capabilities

Protocols

Protocols describe how the repository makes its information, submissions, and entries available: what syntax, semantics, and specifications are supported in the data and commands that go into the system, and come out of it?

Ontology download formats

Ontology upload formats

URI format for concepts and artifacts

API support

How many/which features are available to other applications via API calls? Are these calls REST-based?

How is the URI for each concept in the repository represented? How is the URI for each artifact represented? Are the offered URIs versioned, unversioned, or both?

Usability

Maintainability

  • It must be possible to migrate the content and functionality of the repository to new technologies as they change over time. For example, migrating from google code to github seemlessly...

Ease of content submission

Ease of content updates

Documentation

Responsiveness

Reliability/Uptime

Authentication and Security

What types of login features are supported?

Support for Teams and Roles

Can teams be defined of multiple members? Can multiple members of a team edit the same ontology?

Cost

Costs are given in labor hours and funds required to start and operate the service.

Installation

Operations

Maintenance

Comparative Evaluation of Portal Software

Justifying Semantic Web Repository Business Costs

(see the [SW OPortal Discussion:Talk|Discussion] page for ideas.)

References

[1] ODM2 evaluation of MMI ORR: https://github.com/ODM2/ODM2/blob/07e4e30fc4d6763b8bcae771ffbb7228fc5a3f65/doc/ODM2Docs/concept_controlledvocabs.md