Information Technology Reference
In-Depth Information
Input
Output
Twitter
16:13:00
19:10:30
① Extract human activity
③ Predict missing activity
② Building timeline action network
Fig. 1.
Approach of building timeline action network
we can say that Twitter is becoming the
sensor
of the real world. In other words, we
can collect activities of people in earthquake disaster from Twitter. We can consider
these activities as the collective intelligence. However, sentences retrieved from Twitter
which are more complex than other media texts, are often structurally various, syntac-
tically incorrect, and have many user-defined new words [7].
In emergency situations, it is important to let the computers make a useful recom-
mendation
in time
. This means that the activities should be collected, and represented
in real-time. However, since tweets depend on users' autonomy, it is highly probable
that the users do not post their activities in real-time. Thus, we need to solve the prob-
lem of missing activity data in order to have activity data in real-time. Additionally, to
help the computer understand the meaning of the data, we should build the collective
intelligence based on OWL.
In this paper, as shown in Figure 1, we first use our previous work [7] to automatically
extracts human activities from Twitter. And then, we design a timeline action network
(TiAN) to represent these activities in real-time. Finally, we propose a novel action-
based collaborative filtering, which predicts missing activity data, to complement the
action network. Moreover, with a
combination of collaborative filtering and NLP
, our
method can handle
minority actions
such as successful actions.
The main contributions of our work are summarized as follows:
-
We has successfully designed the TiAN based on OWL in order to represent the
activities in real-time.
-
We also proposed a method that can predict missing activity data to complement
the action network. Moreover, this method can handle
minority actions
.
The remainder of this paper is organized as follows. We first explain how to extract
human activity from Twitter. Secondly, we design TiAN to represent the extracted ac-
tivities. Thirdly, we explain how to predict missing activity data. We then report our
experimental results, and explain how to apply the action network. After considering
related works, this paper concludes with some discussions of the future work.
2
Extraction of Human Activity
In this paper, we define an activity by five attributes namely
actor
,
act
,
object
,
time
and
location
.Andan
action
consists of a combination of
act
with
object
. For example, in
