Latent Social Information in Group Interactions with a Shared Workspace

Shared artifacts, such as drawings and schemas on whiteboards, sticky-notes with ideas on walls, are often created and interacted with during meetings. These shared artifacts a) facilitate the expression of complex fleeting ideas, b) enable collaborators to establish a common ground and validate each othersâ understanding about the context, and c) extend the validity of shared information by making it permanent. By the end of a collaboration session, the shared content denotes the shared knowledge amongst collaborators, which emerged as a result of a recursive process of storage, transformation, and retrieval from an external memory such as a whiteboard. Although these interactions with the artifacts symbolize the important episodes in a group discussion, still the information contained within them has not been much leveraged in collaboration research. Being well assimilated in the established work culture, collaborators do not heed the interactions with the shared artifacts, and therefore the nature of the social information contained within them is latent. However, from a research perspective this information is valuable and can offer insights into a few facets of ongoing group dynamics and processes. This thesis in particular a) identifies and examines the characteristics of the latent social information, b) studies the relationship of this information with different aspects of collaboration, and c) explores the practical utility of this information in collaboration assessment. We start by designing a meeting technology - MeetHub that enables collaborators to share and interact with artifacts in an unconstrained manner over a shared workspace, and allow us to collect fine-grained interactional information. Then we present user studies, where we extract and comprehend the relevant social information from interactions with the artifacts, and analyze its relationship with collaborative processes. Our findings demonstrate that latent social information is significantly correlated with the task outcome, division-of-labor, and the quality of mutual understanding between collaborators. Finally, we present a prediction system based on this social information, capable of alerting the group members about the poorly grounded episodes in real-time, and thus enabling them to regulate their collaborative behavior. The final contribution of this work presents itself as implications towards the dual nature of shared workspace, supporting the creation and sharing of artifacts as well as an assessment tool.

Exploitation de connaissances sémantiques externes dans les représentations vectorielles en recherche documentaire

Florian Seydoux

The work presented in this thesis deals with several problems met in information retrieval (IR), task which one can summarise as identifying, in a collection of "documents", a subset of documents carrying a sought information, i.e.. relevant for a request expressed by a user. In the case of textual documents, to which we limited ourselves within the framework of this thesis, a significant part of the difficulty lies in ambiguity inherent to human languages. The interaction with the user is also approached in our work, by studying a tool enabling a natural language access to a database. Finally, some techniques which permit the visualisation of large collections of documents are also presented. In this document we first of all describe the principal models of IR by highlighting the relations which exist with some manual technics of IR and document retrieval, developed during the past centuries. We present the principle of document indexing, allowing us to represent documents in a multidimensional space, and the use of this representation by a vectorial model. After having reviewed the principal improvements made these last years with vectorial research systems, including the preprocessings of collections, the indexing mechanism and measurements of similarities between documents, we detail some recent usecases of additional semantic resources (semantic dictionaries, thesaurus, networks, ontologies) reported in scientific literature for the indexing task. We then present more in detail the semantic indexing principle of textual documents by using a thesaurus, consisting in integrating in the document's representation space at least part of the informational contents of hierarchical semantic resources. We propose a general framework allowing us to describe and position various possible techniques to carry out the semantic indexing by adapting, if possible, the specificity of the descriptions resulting from the semantic resources to the data to be represented. We use this framework to describe three families of criteria usable for semantic indexing, each one having its own characteristics. For each of these families, we give the specific algorithms allowing the computation of the criteria. The first two families allow us to consider several criteria already known in feature selection. Moreover we show that, unfortunately, many of these criteria are in fact not very effective for the considered task. The third family allows us to introduce a completely new criterion, the Minimum Redundancy Cut criterion (MRC), built on the basis of the information theory and allowing us to obtain index terms having a probability of occurrence in the collection of documents as well balanced as possible. Finally, we treat the case of semantic index independent of the data (statically choosen), allowing a parameterisation of the level of generality of the index terms. Some of the criteria suggested for semantic indexing has been empirically evaluated. To judge their relevance, we used a well known vectorial system (the Smart IR system) and measured the performances of IR obtained with various reference collections. Those collections was indexed on the basis of the studied criterion, by taking into account the strongly structuring semantic relation of hyper/hyponymy ("is-a" relation), given by two different semantic resources. By comparing results obtained with the performances of a traditional indexing (using the lemmas of the words as representation space), we can show on one hand the relevance of the semantic indexings (in RD) and on the other hand the quality of the proposed criterion (MRC). Concerning man-machine interaction, we present a general outline allowing to build in a relatively fast and systematic way systems with mixed initiative, giving the human user a large (and natural) latitude in the control of the dialogue. This outline is usable in typical database research-task applications (where the database is hidden to the user, but the latter knows exactly which information they wish to find) as well as advice-task applications, for which the users does not necessarily have a precise idea of their needs, and uses the system not only for specifing their wishes, but also a set of propositions as a final result. We particularly stress the techniques allowing us to obtain a robust system, able to deal with speech recognizer failures. Concerning the visualisation of large textual data collections, we present an application of the correspondences analysis (allowing to highlight similarities and oppositions for various groups of entity, built on the basis of additional features present in the DB) to the case of patents data. In addition, we propose a method (based on the bootstrap replication principle) allowing us to determine a confidence interval for relative positionings of various groups, thus permit to immediately judge the reliability of the visually apparent similarities or oppositions.

EPFL2006

Interaction Mirrors for Face-to-Face Meetings

Flaviu Roman

To be successful, a group requires an appropriate quality of coordination among its members. Even with the advent of remote communication, face-to-face meetings remain the context that offers the best interaction support to an effective collaboration. However, face-to-face meetings seem to suffer from recurring shortcomings which are yet to be alleviated. One of them is the omniscience and persistence of undetected gaps between the actual and the desired (or intended) state of the meeting. The evolution of technology has spawned the creation and availability of large sets of data, giving birth to what we call today analytics, the science of discovery and communication of meaningful patterns in this data. This thesis aims to explore the potential of technology to process meeting-related data and help reveal these gaps to the participants. We also conduct an analysis of the actions that the participants perform, as an effect of becoming aware of such gaps, and we reflect on their possible role in increasing the effectiveness of the meeting. Our technologies collect interaction data in real-time during face-to-face meetings, perform analytics to transform this raw data into explicit representations, and then use visualizations to communicate knowledge about the interaction back to the participants. This thesis expands the concept of Group Mirrors, to define the concept of Interaction Mirrors, which are tools that perform this instant transformation of meeting data into explicit repre- sentations, and return this knowledge back to the group during the meeting. We present the implementation of three types of Interaction Mirrors that tackle aspects of interaction related to Participation, Time management, and Mutual Knowledge. They vary with respect to data collection, being either implicit (continuous) or explicit (discrete), and to their presentation to the group (continuous or discrete). We perform a total of 7 experiments, ranging from basic usability observations, to year-long, longitudinal studies. Conducted in real-world situations, most of these studies are set up in the field, in various organizations, corporations and institutions. We show that each type of interaction mirror has an influence on the group, and discuss the capacity of this influence to improve the outcomes of the collaboration.

EPFL2013

Knowledge interdependence with the partner, accuracy of mutual knowledge model and computer-supported collaborative learning

Pierre Dillenbourg, Gaëlle Molinari, Mirweis Sangin

The present study is part of a project aiming at empirically investigating the process of modeling the partner’s knowledge (Mutual Knowledge Modeling or MKM) in Computer-Supported Collaborative Learning (CSCL) settings. In this study, a macro-collaborative script was used to produce knowledge interdependence (KI) among colearners by providing them with different but complementary information. Prior to collaboration, two students read the same text in the “Same Information” (SI) condition while each of them read one of two complementary texts in the “Complementary Information” (CI) condition. After the collaboration phase, a knowledge modeling questionnaire asked participants to estimate both their own — and their partner’s outcome knowledge thanks to Likert-type scales. The relation between the accuracy with which co-learners assess their partner’s knowledge and learning has been examined. In addition, we investigated the KI effect on (a) learning performance and (b) the MKM accuracy. Finally, we wondered to what extent the MKM accuracy could mediate the KI effect on learning. Results showed no difference in learning performance between participants who worked on same information and participants who worked on complementary information. We also found that participants were more accurate at assessing their partner’s knowledge in the SI condition than in the CI condition. The discussion focuses on methodological limitations and provides new directions for investigating the KI effect on MKM accuracy.