Java et al. (2007) present a topology of various uses
and usage intentions on Twitter. They further find
that Twitter’s users extend the Twitter question,
‘What are you doing?’ by including information
sharing (e.g. URLs) and normal conversations on the
platform. Further statistical studies of Twitter were
presented by a number of researchers including
Krishnamurthy et al. (2008, identification of
different user types) and Huberman, Romero and
Wu (2009, networks on Twitter).
Going one step further, some research has been
conducted to develop behaviouristic models for
explaining the ‘why’ in microblogging (Barnes and
Böhringer 2009, Günther et al. 2009). However,
with only two existing works on modelling
microblogging behaviour, there is no established
research framework yet. Clearly, there is a need to
understand the underlying principles of
microblogging services like Twitter in order to
design and implement successful expansions like
ubiquitous microblogging.
R1: Build and test predictive models for user
behaviour in microblogging services.
Enterprise adoption of microblogging is in a very
early stage of development. Therefore, there are not
yet many users in the enterprise context, which
would be a prerequisite for broader studies of the
area. As an alternative, existing cases should be
investigated in detail. Hence, the case study method
is a reasonable research methodology for learning
about microblogging in enterprises. To the best of
our knowledge there exist only two research case
studies (Böhringer & Richter 2009, Barnes et al.
2010), yet. Collecting and analysing more cases and
conducting multi-case analyses are important future
research tasks.
R2: Conduct case studies and multi-case
analyses.
4.2 Technical Issues
There is significant potential in supporting users in
searching the ‘hair stack’ for important information
(objective O9). For Twitter there exist a number of
small applications for visualisation and text mining.
However, they are prototype implementations and
are not integrated to each other or well-documented.
From a scientific point of view, little research has
been done in this area. Only Assogba and Donath
(2009) argue for a stronger visual support of
microblogging users and present a platform for
‘visual microblogging’.
Further, the user would benefit from support by
automated agents understanding the semantic
meaning of the information. For this reason, Passant
et al. (2008) present a concept for semantic
microblogging. However, the main question for the
enrichment of text with semantic information is user
acceptance, which is an important point in
microblogging as a medium that is mainly based on
the ability to publish information fast and easily. An
alternative to semantic markup of microblogging
postings is to understand the meanings of natural
language using Natural Language Processing (NLP).
However, it can be questioned how effectively NLP
could work on the very short, often informal and
possibly flawed microblogging postings. First
experimental prototypes have demonstrated the basic
feasibility of the approach (see, for example, the
service akibot.com).
R3: Support users with visualisation and
analysing functionality including semantic
richness.
4.3 Conceptual Issues
Our approach of ubiquitous microblogging implies
challenging conceptual issues. An important feature
of microblogging is the chance to interact with other
microbloggers. However, in a microblogging space
with actors like machines, sensors and processes,
this is hard to achieve (objective O4 conflicts with
O6). A user might want to ask ‘@machineR2D2
when is your next maintenance due?’. As discussed
above, text mining/NLP could be a possible solution,
but it is questionable whether the technology is
mature enough to achieve a good reliability. A
solution could be to divide microblogs in
unidirectional and bidirectional blogs. Unidirectional
blogs like @machineR2D2 could be associated to a
bidirectional channel, like the microblog of the
machine’s technician, for providing a backchannel.
The research challenge here, again, is solving these
problems with respect to simplicity and transparency
for users.
Another level of complexity evolves due to the
usage of mashups. Mashups could be very useful for
fulfilling O8. However, if mashups re-arrange
postings and are published as new microblogs, the
same posting exists twice. For example, a mashup
could aggregate all the microblogs of a machine
park’s devices and filter for the keyword ‘error’. The
result of this mashup could be a new microblog
‘@machine_errors’. Problematic in this scenario is
that the same event occurs in two microblogs, which
could lead to distorted analyses and double
ICEIS 2010 - 12th International Conference on Enterprise Information Systems
68