View source for 18.05.2012 WP10-WP5 - X-Search Integration

==Agenda==
Time: Fri, May 18, 2012, 11:00 - 12:00 CET

*X-Search integration with gCube.
*Modeling of resources needed by X-Search in gCube.

==Participants==
*Leonardo Candela (CNR)
*Andrea Manzi (CERN)
*Yannis Marketakis (FORTH)
*Yannis Tzitzikas (FORTH)
*Gerasimos Farantatos (NKUA)
*Rena Tsantouli (NKUA)

==Discussion Summary==

X-Search is used on top of an underlying search engine and provides a number of semantic search features. It is composed of components implementing these areas of functionality. 

FORTH has provided a document including suggestions and issues that need to be discussed for the integration of X-Search with gCube.

FORTH described their suggestion for committing and packaging the components comprising X-Search. 

FORTH will start committing code to SVN, under a folder dedicated to semantic search and use maven to build the components. ETICS component configurations will be created under the new subsystem for semantic search.

FORTH asked if maven should be used to build the new components. CERN informed FORTH about the decisions made regarding the usage of maven, i.e. that new components should prefer it. FORTH agreed to use maven.


FORTH currently develops a portlet which will be used to interface with X-Search. FORTH asked which is the difference between using the ASL HTTP API and using a dedicated client in terms of efficiency. 

NKUA suggested using a client such as ASL Search as the most efficient method. However, ASL Search cannot be used if X-Search is going to be deployed as a service, since the former must be collocated with the portal. A possible solution would be for X-Search components to be deployed as libraries at the portal node and used by the portlet. This
however may be not possible due to the requirements in terms of computational power of the latter. In that case, a different scheme should be used.

NKUA and FORTH will investigate computational requirements further, in collaboration with people responsible for portal functionality.

FORTH asked if using a client will have impact on access to snippets. NKUA informed that the mechanism is transparent and case result consumption is made through gRS2, the snippet will be present as a record field.

CERN suggested that the deployment scheme be clarified, and reflected in the specifications of X-Search.
FORTH will replace the component diagram in [https://gcube.wiki.gcube-system.org/gcube/index.php/Semantic_Data_Analysis MS45] with a proper deployment diagram as soon as decisions are made.


FORTH described the work done in collaboration with FAO in the direction of exploiting the FIGIS search engine. The target of the integration of X-Search with gCube should be to be able to achieve exactly the same functionality that is provided by X-Search over FIGIS. FORTH in addition poses the question of whether FIGIS should be integrated into the gCube Search System itself, thereby allowing X-Search to exploit the distributed nature of the former and be able to obtain results from a single endpoint.

NKUA described the procedure of integrating a new Data Source into the gCube Search System and asked if FIGIS can be modeled as an OpenSearch data source.

FORTH informed that FIGIS does not expose a description document, but is almost compliant with OpenSearch. 

NKUA replied that in that case no significant effort is needed, we however need to better understand if the inclusion of new sources
is justified and which the benefits that will be gained are.

CNR pointed out that a more concrete plan is needed and that effort should prioritize requirements originating from the CoP. In order for the entire consortium to be aware and contribute to the activity,a ticket to monitor all activities on X-Search should be opened. FORTH agreed to create the ticket.


FORTH described the usage of SPARQL endpoints by X-Search and asked for information about the modeling of such endpoints in gCube. 

CERN and CNR explain the modeling of resources in gCube, CNR informs that SPARQL endpoints, regardless of being internal or external to the infrastructure, are resources that need to be discovered and used. They therefore need to be registered to the information system. The discussion is timely because of the current focus on the new resource model by WP8. 

FORTH and NKUA are going to collaborate with WP8 to decide how SPARQL endpoints are going to be included in the resource model.


FORTH mentioned a problem regarding differences in appearance between the portlet they are developing and existing gCube portlets. 

NKUA informed FORTH that the correct styles are used when the portlet is deployed in gCube portal and about the usage of gCube widgets.

NKUA will assist FORTH further if other issues of such kind are encountered.

==Actions==
*FORTH will start committing components to SVN, under trunk/semantic-search.
*FORTH will open a ticket for the task of X-Search integration with gCube so that the activity is better coordinated and visible throughout the consortium.
*FORTH and NKUA will collaborate with WP8 in order to see where resources exploited by X-Search fit into the resource model.
*NKUA portal team to investigate the requirements of the new portlet with respect to computational power in collaboration with FORTH and CNR, in order to aid in deciding the final deployment scheme of X-Search.