In 2013 a Task Force for each COSCH Working Group was formed. A series of working meetings has since been organised to enable in-depth discussion of key scientific questions concerning spatial and spectral documentation of material cultural heritage. Alain Trémeau and Orla Murphy report on the first meeting of the Task Force for the Working Group 3 whose work is centred around three questions: registration processes, integration of multi-sensor data, and data access and formats. The meeting was held at the Universitat Politècnica de València, Spain, on 30th and 31st January 2014 and focused on this Group's contribution to the COSCH Knowledge Representation under development.
Report of WG 3 Task force meeting in Valencia, Spain
30th–31st January 2014
The main focus of the meeting at Valencia was a discussion of the COSCH Knowledge Representation (KR). Initially the acknowledgement of the gap between the expectations of technical people and users of the knowledge representation model was explored in a detailed discussion. Key questions were articulated in order to assess the perception of all users regarding KRM as a useful, reliable tool for knowledge sharing including:
What results do users want to get from this model?
What is their knowledge domain?
What are their activities within their domain?
What are the key concepts that are critical for users?
What information do users need?
The potential success of the KRM rests in the acceptance of groups of users that their expertise may be captured in a global knowledge base, the concept of a ‘data story' was considered as a means of bridging knowledge gaps in this regard.
The issue of vocabulary, a central issue for the KRM was discussed and parsed in depth. Across the domains the term ‘ontology' is most appropriate, so that we should endeavour to build a bridge between perspectives – semantic web, knowledge representation, ontology and knowledge inference. This is regarded as a crucial element for the success of the KRM. Initially the KRM cannot cover all technologies and cannot be exhaustive – instead it will be progressively enriched. This is an organic approach to knowledge representation and design facilitated by semantic web technologies. In previous database types both hierarchical and relational, this type of data structure was fixed, not so with this knowledge schema approach.
Several knowledge domains are represented within the action – it is clear that some of these must be at the top level of representation within the knowledge model: objects, technologies and instruments. Others must be transparent to users: data type, algorithm. For some applications cost, time and technical constraints are the main parameters for consideration. This tool should provide the information that the user is looking for and must guide the user in order to provide the information most useful for them in a given circumstance. As an example, the user may not need any information about algorithms at the beginning of their query but should be aware after some steps that the accuracy of the instruments proposed by the system depends on some parameters including those of the algorithm used. Consequently, we must provide them with some information concerning the selection of an algorithm among a class of algorithms if required. This transparency is vital for academic rigour as within such a context ‘black-boxing', or the inability to interrogate or parse a technology in context, is to be avoided when possible. Researchers are users who need to thoroughly explicate their process in terms of their research.
The workgroup proposed a series of examples or study cases to explicate the key ideas within the knowledge representation model showcasing it as both flexible and responsive to the needs of a wide range of users, in a range of circumstances, from naïve user to those with expert knowledge. A study case pertinent to each work group was suggested, and Prof. Treméau and Dr Karaszewski agreed to build these initial examples around a spectral and UV solution, and a 3D and colour solution respectively.
There was further discussion on potential H2020 bids organised within the group, with a number of ideas clearly articulated – subsequently these have been further developed through collaboration between researchers and practitioners across the COSCH network. The first call deadline is 30th September 2014. Potential ITNs Innovative Training Networks were also discussed within the context of the MCSA element of H2020. It was agreed to remain in contact with NCPs (national contact points) in order to further the aims of this inter and transdisciplinary group within broader EU development and funding programming.
The discussion turned to documentation of algorithms and procedures and the ASM the algorithm selection module. Again the discussion centred around how this may be used to combine domain expert knowledge, for example semantic data about geometry, the parameterization process and algorithm knowledge for example algorithm classes for each image processing task, and semantic rules – how in effect, may we put the user in the loop and involve them actively in the process.
An incremental approach was decided upon:
1. Consider the relevance of the framework – from the point of view of object, data, and the technological point of view
2. Consider the validity of the structure through the use of examples – for example on colour calibration, on 3D calibration and reconstruction. Prof. Trémeau successfully prepared an example of colour calibration (algorithm behavior) and presented it the Joensuu meeting. Dr Gerke will prepare another example of structure from motion.
3. Assessment of user satisfaction and engagement, moving towards a final concept.
It was agreed to submit a joint paper to the midterm symposium of ISPRS (23-25 June at Riva Italy) and to hold the next work group meeting there. This paper has subsequently been written, accepted and will be published in the proceedings of the conference. COSCH will present a panel of papers at ISPRS again this will be a transdomain initiative focusing on the active engagement across the group.
The focus throughout the two day task force meeting was on an integrated approach to the problems of scattered, unfocussed cultural heritage solutions in the EU. Users' solutions remain at the centre of concerns for a semantically enabled solution for the EU's Cultural Heritage particularly those solutions addressing spatial and colour concerns.
Alain Trémeau and Orla Murphy, Chairs of COSCH Working Group 3