The AXIS Conceptual-Reference-Model (Concepts and Implementation)

Introduction to the AXIS-CRM and to its implementation
1. The AXIS-Conceptual Reference Model
1.1 Generalities
A new modular and tailorable approach for the semantic modeling of static and dynamic knowledge has been elaborated under the name “AXIS Conceptual Reference Model” (AXIS-CRM). AXIS organizes that modeling as networks of Autonomous Semantic Objects (ASO). In turn, each ASO is expressed as a network of Elementary Semantic Entities (ESE). The ASO wraps the instances and their models to becomes ‘autonomous’. At Elementary Semantic Entity level (simply called ‘Entities’) the modeling uses four leveled constructors: Term; Document; Relation; Profile. The knowledge models and their instances are represented by a collection of Documents (among with the OWL files expressing the models) bundled by a Configuration Management Document (based on RDF). These collections are semantic Entities that can represent any topical subject or object. These Entities are linked through typed Relations. The dynamic aspects (events) and the imports / exports are also managed by dedicated Entities.
AXIS-CRM permits the covering of any semantic domain. One of its innovative contributions is the introduction of the “AXIS-CORE” ontology and profile (see below) and of an “Ontology Plug-in” mechanism required for addressing specific domains, whether in economic, cultural or scientific assets. The methodology has been tested by derivation of existing ontologies such as the FOAF and IAML. Native new profiles have been developed, in particular to cover the Audio-Visual Interviews and the News network according to the NewsML-G2 format. The authoring of these specific ontologies is made according to a strict procedure involving experts of the domain and a four step process (Terms; Relations; Documents; Profile) assisted by a “Knowledge Modelling Tool”. At the semantic level, the AXIS-Core presents what the Dublin-Core has presented at the Term level. The AXIS-Core could be used as seed for easy extensions according to specific needs although keeping the interoperability. The approach is ideal for distributed operations, collaborative work and evolution / migrations.
The approach is totally fitting in the SNIA model of acquisition, operations and fixing of the knowledge. Furthermore, it resolves the famous opposition between ‘Unstructured’ ↔ ‘Structured’. In the current SNIA model, in ‘Structured’ the data are represented in databases; while in ‘Unstructured’ the data are represented as documents conforming to external formats (standard or proprietary). When the formats of the documents and database-records are explicitly defined in semantics models, both representations can be used and converted to each other. Further, the SNIA model can be extended to the “LIVE”! The extended SNIA model becomes:
Figure 1: The extended SNIA model
1.2 The ENTITY Constructs (Models & Instances)
The Entities are represented according to a uniform modelling technology which covers as well the assets, the work-flows, the presentations and the agreements.
The model constructing and validating the instances is called PROFILE. It includes six components covered in a multi-linguistic way.
2. The URN system
3. A set of specific ‘TERMS’ classes
4. A set of specific ‘DOCUMENT’ classes
5. A set of specific ‘Lists of authority properties/values’
6. A set of specific ‘GUI frames’
The classes defined in the AXIS-Core profile is illustrated by the figure below:
 Figure 2: AXIS Core Classes 
The static instances are constructed according to the applicable assembly of the Core profile with the domain specific profile: ‘Music’ (derived from the IAML ontology proposal). The example of the instances and class associations is illustrated according to the figures below:
Figure 3: Example of application at instance level
The dynamic instances (work-flows) are constructed according to the applicable assembly of the Core profile with the domain specific profile: ‘Finite-State-Machine”. The example of the instances and class associations is illustrated according to the figures below as a chaining of elementary processes controlled by state vectors.
Figure 4: Example of Work-Flow
1.3 The Autonomous Semantic Objects (ASO)
The ASO are represented according to a uniform modelling technology which covers (at model and instance levels) a semantically coherent network set of assets, work-flows, presentations and agreements.
The ASO are exported from a “Local Operation Centre” managed by an instance of a special class of Entity called “Autonomous Resource Entity’ (ARE). They can be imported to another ARE or converted to another semantic format or migrated to non semantic formats (like MXF or METS or MPEG-21).
 Figure 5: Example of Network of ARE exchanging ASO
1.4 Ownership of intellectual rights
AXIS-Conceptual Reference Model is owned by the Non Profit Association TITAN:
Contacts: M. Roger Roberts (President) [email protected] M. Guy Maréchal (Administrator) [email protected]
2. The implementation of the AXIS-Conceptual Reference Model
AXIS-CRM has been implemented by a demonstrator and extensively evaluated in three challenging applications:
•   Migration of existing flat models and associated media to ASE
•   Open Interchange of ASE between heterogeneous systems
•   Creation of native persistent archives conformed to the OAIS ISO standard.
The demonstrator has been made using standard languages (mainly OWL, RDF, METS, MXF and SMIL) and on standard reference models (mainly OAIS, FRBR and CRM). The evaluations have been conducted by extensive user’s tests and scientific analysis. In particular, it has been demonstrated that AXIS is a superset of FRBR-oo.
The implementation has been made in two European Projects (MEMORIES and MediaMap) makes use of existing achievements, in particular from European Union projects (PrestoPrime, Europeana, Film-Gateway, Video-Active, AceMedia …) and is largely based on solid open-source software.
The demonstrator includes a “Semantic Profile Modeling System”, a “Semantic Authoring System” (called IPI-Manager) interacting with a “Semantic database” (called ISIS) and in a generator of “Autonomous Semantic Objects” (ASO). The implementation has been made by MEMNON.
The focus is on the opening (public availability under open licence) of the formats and protocols of the import/export of the systems (the ASO).
The exploitations opportunities of the AXIS modelling of the assets are numerous. One of the most important one is the management of the interoperability. Three main interoperability targets are in sight and a proof of concept of key elements has been implemented in the MediaMap project:
o  The migration of data models. Most of the current information objects are currently represented by “Flat” models usually as record in a relational data base. The trials have confirmed the capabilities of migration of these ‘Flat’ models to ‘Semantic’ models by a two steps process. First ‘enhancement’ with alias; then ‘enrichment’ by the addition of the semantic data and the insertion of the Information Objects in the semantic network.
o The interchange of Information Objects between systems. When Information Objects are represented in a semantic way and that the document modeling technologies and formats are explicitly available, it is possible of implementing “Inter-Operability Wickets” which could operate in a quasi full automatic way. This allows the interoperability between independent systems because the meaning of the data can as well be understood by the human but also by the IT machines.
o The preservation of Information Objects. The Information Technologies are permanently changing! The data representations of the Information Objects become then quickly obsolete. The preservation of the value and the access to the Information Objects is essential. Here the interoperability solution is simply the same as the previous one. The interoperability of the representations between their representation formats valid a few years ago and those according to the current today standards or practices. It is the interoperability in time which paves the way to the persistence! It means that the assets have to include constructs that implies their capacity of evolving according to the properties expressed for the persistence, in particular in the ISO standard OAIS.
Ownership of product rights: The implementation of the AXIS-CRM is owned by MEMNON: Contacts: Michel Merten (President) [email protected] Guy Maréchal (Strategic adviser) [email protected]
3. Open Availability of the AXIS-Conceptual Reference Model & Tool kit
Mid 2010, the AXIS Conceptual Reference Model [AXIS-CRM] and its associated Tool-Kit will be made openly available on the UNESCO open repository of the “Memory of the World” program.

About the Author(s)