ICMI 2005 Workshop.
The interest in automatic collection and analysis of
meeting recordings is constantly increasing in the research community.
Current efforts try to focus not only on the single modality of
speech, but to take a broader view attempting to derive useful
information from meetings, based on multimodal perception and
understanding of a wide array of information sources (gesture,
handwriting, sketches and other manual activity, body and head pose,
eye gaze, email leading up to a given meeting, documents that are part
of the subject matter or background for a meeting, agendas, lists of
critical outcomes, etc.). Such a wide spectrum of input sources gives
the opportunity to explore truly multimodal processing approaches,
which remain a hard and open challenge under many aspects. Technology
has still to be proven effective as a mean of handling meetings, from
both offline and online perspectives.
Offline processing
technologies are aimed at making meeting recording archives a valuable
asset, by extracting the content of meeting archives and capitalizing
on the knowledge they contain. Important research efforts are directed
towards more and more complex content analysis algorithms producing
useful indexing material, ranging from fact detection and extraction to
analysis tasks involving higher level interpretation, such as
participant interaction analysis (is there agreement ?) or evaluation
of the meeting development (has any decision been made ? what is the
agenda ? did the meeting reach the initial goals ?). At the same time,
sophisticated interfaces moving beyond simple content reproduction and
allowing users to access and use effectively data of such a complexity
must be designed.
Online processing is aimed essentially at
developing systems that support colocated meeting participants
activities and/or involve remote participants. In such a situation,
computers become the channel of human-human interaction and represent a
bottleneck resulting into non-natural feelings as well as lack of
communication effectiveness. For this reason, many researchers have
studied ways to use computational support to create collaborative
environments that is at least good as, if not better than, "being
there". The use of multimodal interfaces can address the problem by
conveying more information and/or driving the attention of the users
towards actually important elements.
The goal of this workshop is to
gather researchers and technology developpers of all horizons, active
in the above or related domains, in order to acquire a broad view of
current state-of-the-art, share experiences, exchange ideas and
establish collaborations and contacts. The workshop will be the place to
discuss the opportunities and effective usefulness of newly developped
technologies for meeting applications. Thus, we are looking for
position as well as research papers debating on or contributing to the
following (and other related) areas:
- Smart meeting rooms, Meeting data collection and Annotation tools
- Multichannel processing
- Multimodal identification of intent and emotion
- Multimodal person identification
- Meeting dynamics and human-human interaction modeling
- Multimodal dialogue modeling
- Remote collaboration in meetings
- Content abstraction, summarization and structuring
- Multimodal indexing and retrieval