Mulce.org Documentation

Analysing different types of interaction data with SAMSA, a Social Network Analysis tool

1) A short description of the pedagogical context:

SAMSA is a tool that enables evaluators to perform social network analysis of data representing interactions in collaborative settings. Due to the nature of the analysis, we focus our research in settings where participatory aspects of learning can emerge. Normally, this means that at least three people have to be involved in the collaboration, although the analysis will be more appropriate for larger groups or communities.

2) A short description of the corpus extract:

One of the main ideas that drove the desing of SAMSA was that it could be applied to different types of interaction data. At this moment, it can be applied to different corpora representing different types of interaction.
We plan to provide a corpus coming from two courses supported by BSCW (a shared repository of objects developed by Fraunhofer in Germany). A more detailed description of the corpus will be provided at the workshop.

3) A short description of the format used by the tool:

The tool uses the so-called "common data format", developed by a group of research teams within the Kaleidoscope Network of Excellence. Its structure can be viewed in the following figure:

4) A short description of the downloadable tool:

SAMSA is an stand-alone tool that allows evaluators (and eventually, teachers) to perform social-network analysis on different types of interaction data, from direct interactions from dialogue-based interfaces (such as e-mail, fora, chats) to object-based interfaces where collaboration is mediated by objects (such as the mentioned BSCW or other tools based on the manipulation of objects in a shared space). The requisite is that these interactions comply with the mentioned "common-data format".

SAMSA is able to load data from the context of the interactions from the files, but it also allows the evaluator to configure it from the interface, adding new users, groups, etc.

SAMSA allows to configure the type of network that is to be analyzed, with a number of parameters, such as the dates to be considered, the type of network out of three pre-defined types of interaction (direct, indirect, object-based), and also, the kinds of actions that will be taken into account to define a relationship in the network. For example, an evaluator can be interested in the relationships between an actor that creates an object and those that modify that object; or in the relationships between the creator and the actors that read that object, etc. The evaluator can also configure the SNA-indexes that wants to be computed among a set of indexes provided by the tool.

From these configuration parameters SAMSA builds the network, and computes the indexes chosen by the evaluator. The system provides different types of output: a graph with the sociogram; tables with the actors and their indexes; and files that can be used by standard SNA-tools to further exploit the results.

SAMSA has been developed though a series of incremental prototypes, where new functionalities were added on top of the existing ones, and which made it rather unstable. It is currently under a process of re-ingeniering. A new release is expected for the end of August, and will be provided at the workshop.

Web-site for our SAMSA tool where it can be downloaded (a beta version): http://www.gsic.uva.es/samsa.

5) A short description of the research questions associated to the tool:

The research questions associated to the tool are, for example: how dense have been the interactions among actors in the whole group? or where are the different actors situated into the network? (are they central nodes, or, on the contrary are at the periphery?), how cohesive has been a group or sub-group in the network? what can we say of the differen roles that can be detected? etc.

Return to the detailed programme of the workshop: EUROCALL 2010, workshop "Dissemination and comparison of research findings: developing Contextualized Learning and Teaching Corpora (LETEC)"

References:

In the last CSCL we presented a position paper with our ideas regarding
adaptivity and interoperability of the IA systems.

The reference is:
Alejandra Martínez-Monés, Jose Antonio Marcos-García, Mª Jesús
Rodríguez-Triana, David Fernández-San Miguel
Proposals for adaptive and interoperable IA systems, CSCL 2009 workshop on "Interaction Analysis and Visualization: Analysis’ Methods, Tools and Research Questions", Available on:
http://www.ltee.gr/IA_WS_CSCL2009/CSCL2009_WSBruillard_martinezetal.pdf.

The reflections regarding the difficulties to integrate collaboration
monitoring and analysis functions in CSCL have been discussed in a chapter:

*Alejandra Martínez-Monés*, Andreas Harrer and Yannis Dimitriadis, "An
Interaction-Aware Design Process for the Integration of Interaction
Analysis into Mainstream CSCL Practices”, in Sadhana Puntanbekar,
Gijsbert Erkens, and Cindy Hmelo-Silver (eds), Analyzing interactions in
CSCL: Methodologies, Approaches and Issues, Springer, 2010 (in press)

Finally, for a more complete approach to our proposal of a mixed
evaluation method supported by SAMSA, with an example of application of
the tool, the paper at Computers and Education, although a little bit
old (2003) is still valid:

Martínez, A., Y. Dimitriadis, B. Rubia, E. Gómez and P. de la Fuente.
Combining qualitative evaluation and social network analysis for the
study of classroom social interactions. Computers and Education 41(4):
355-368, 2003.
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6VCJ-49MDYG3-1&_user=1517318&_coverDate=12%2F31%2F2003&_alid=1406988698&_rdoc=1&_fmt=high&_orig=search&_cdi=5956&_sort=r&_docanchor=&view=c&_ct=3&_acct=C000053451&_version=1&_urlVersion=0&_userid=1517318&md5=12ca4192290c225e5e2557470380c959


Home | Contact | Site Map | | icone statistiques visites | info visites 32379

Follow site activity en  Follow site activity News  Follow site activity Eurocall Mulce Workshop   ?

Site powered by SPIP 3.1.3 + AHUNTSIC

Creative Commons License