On Using Social Context to Model Information Retrieval and Collaboration in Scientific Research Community - Community-Built Databases: Research and Development - page 137

6.2.2 On Using Social Context in Information Retrieval:
Toward Social Information Retrieval
Nowadays, it is well known that traditional retrieval models make information
retrieval difficult and challenging from the cognitive side, particularly in large-scale
and interactive environments supporting communities such as bloggers, Wikipedia
authors and users, online communities through Facebook, Myspace, and Skyblog.
The main criticism is that, in these approaches, retrieval ignores the influence of
user's interactions within his social context on the whole information process. Thus,
the use of social networks' theoretical foundations becomes tractable to achieve
several retrieval tasks. Accordingly, social information retrieval [ 18 , 19 ] became a
novel research area that bridges information retrieval and social networks analysis
to enhance traditional information models by means of social usage of information.
Social context is inferred from the analysis of unstructured communication
between users [ 20 ]. It refers to both user profiles, interaction with each other, and
tasks achieved through sharable information spaces, mainly Web sphere. Indeed,
analyzing what people exchange, say, share, annotate, etc., allows to be identified
so as to better meeting their information needs.
In practice, the social context includes personal context and implicit and explicit
indicators of interest and information usage such as tagging, rating, and browsing
activities, communication with friendships, and activities of friends. [ 21 ]. Accord-
ing to the main objective achieved, we can categorize state-of-the-art works on
using social context for information and/or knowledge management and retrieval in
general, as follows:
l Online communities identification [ 22 , 23 ].
l Social recommendation and filtering [ 24 , 25 ].
l Collaborative information production and sharing [ 26 , 27 ].
l Folksonomy and tag prediction [ 28 , 29 ].
l Expert search [ 30 ], people search [ 31 ].
l Social information retrieval [ 32 , 33 ].
In particular, social information retrieval, which is the subarea addressed in this
chapter, consists mainly of using the social context during the document retrieval
process. The main challenge of social information retrieval is, thus, to use social
metadata to cluster the user's information needs according to his socially close
neighbors and consequently adapt the relevance assumption of documents. With
this in mind, our objective here is to reach a wide definition of context in a social
dimension involving users seeking scientific information. Thus, key components
such as information, information producers (authors), information consumers (sci-
entific users), and several explicit and implicit relationships between them such as
production, retrieval, interest sharing, authoring, trust, citation, and bookmarking
are modeled and used as clues for estimating document relevance according to
user's information needs. In what follows, we first review prior works on social
information retrieval within a scientific community and then outline our main
research contributions.