Conclusion about the use of SOA and recommendations

From Stellar Deliverable 6.4

From a research perspective, the idea to have in simple way access to the newest and most complete bibliographical information, including online access to the resources, is surely supported by the community. Moreover, either for PhD students (e.g. doctoral consortia), specific research groups (e.g. Theme team) or events (e.g. Alpine Rendez-vous), there is a need for tools allowing setting up a professional environment efficient and time saving to support collaboration, exchanges and visibility. The Stellar scientific portal, based on the OA, is designed to fulfil these expectations, however it meets two types of concerns related to the rapid evolution of the internet and of the practices. On the one hand, there is a large offer on the Internet of solutions for individuals and groups to create a large variety of tools to share, exchange, disseminate, and relate resources and scientific outcomes. These solutions are not content specific and rather fragmented, but the easiness of their access and of their use is seducing. Let us here quote a scenario (based on a real case) we wrote for the deliverable 6.3:

Let us look for the TEL literature on personalisation, motivation and authenticity. Using Google Scholar (and limiting it to the last five years), a very big number of references is obtained that make impossible a relevant use of them (79600 for this set of keywords). Moreover, the source on which Scholar makes its search is so general, ill defined and badly standardized, that it is very unlikely that it can help at this stage. Actually, one thing is forgotten:

The nature and quality of the resource you are exploring is as important as the preciseness of your request. This is the baseline of the design of SOA and of the TeLearn OA to which it is related. Hence, to pursue the development of a specialised OA is confirmed, but this can be done only if researchers who are both providers and users can contribute to it at the lowest cost.

In order to decrease the cost and effort for populating SOA it has been decided to create automated processes to get the bibliographical information, treat it and make it available. This decision was based on a common feeling that all the institutions were able to provide a feed for this purpose. It appeared soon that these feeds did not exist in all cases, and that their content and format were rather heterogeneous—even in a small network like Stellar. Then, it has been decided to limit the bibliographical description to a minimum information (author, date, title, localisation) for an automatic treatment, but to leave open the possibility to update this description when the researchers are interested (for example adding key words or an abstract). This minimum is rather poor, but it allows still SOA to be OAI-PMH compliant and does not prevent it to improve its quality once it has passed the threshold of several thousands of references which is necessary to be seen as a significant resource for the community.

In this context, SOA and TeLearn have clear complementarities. SOA will make available a large data base of TEL specific references but with a minimal documentation, TeLearn will continue to develop as a high quality repository facilitating thin and scientifically rich literature search, this evolution is already visible in the current status of both as the following figures show:

Number of resources	SOA (704)	TeLearn (1364)
single bibliographical reference (notice)	10 (1.4%)	38 ()2%
notice plus link to the document	683 (97%)	156 (12%)
notice plus associated document (pdf)	11 (1.5%)	1170 (86%)
abstract	27%	100%
affiliations	9.5%	100%
number of videos	1	120

Comparison of SOA and TeLearn stores

Although the priority is given to gathering the information on the scientific production of Stellar members, SOA will not be limited to these resources. It will also be open to references from conferences and journals, as well as other repository within the TEL research area. For this purpose, SOA will become 'harvester', by coping with the feeds issued from the various institutions and sources. SOA will also make available and API to facilitate the dissemination of its content. This API will be released at M24 when we expect to have a consolidated repository which content will be minimal but reliable (no wrong information, no unwelcome replications, etc). At the present time a widget has been created to allow partners sites, and in particular Teleurope.eu to get and display the newest information about the content of SOA.

This effort to provide Stellar, and the community beyond Stellar, with such a resource as SOA has already an integration impact: sharing a strategy as well as the resources, agreeing on the description of the items stored. The evolution which is expected towards a better description of the resources after the bootstrapping period, must leverage the attention to the other aspect which is that of the scientific interactions on top of the repository. Tools are there, to encourage their use and demonstrate the benefit is an other issue which should be discussed and to which it should be responded at the level of the Stellar instruments. SOA will contribute to this effort by the means of the network of the SOA facilitators, with a special effort in targeting PhD students and Stellar events. The collaboration with Teleurope.eu will enlarge the scope of the public reached, while keeping the scientific and focussed objective of the Stellar Scientific Portal.