Text-based Event Temporal Resolution and
Reasoning for Information Analytics in Big Data
1
Junsheng Zhang,
1
Changqing Yao,
1
Peng Qu,
2
Yunchuan Sun
1
Institute for Scientific and Technical Information of China, Beijing 100038, China
2
Business School, Beijing Normal University, Beijing, 100875, China
Email: {zhangjs,yaocq,qup}@istic.ac.cn yunch@bnu.edu.cn
Abstract—Events formulate the world of human being and
could be regarded as the semantic units in different granularities
for information organization. Extracting events and temporal
information from texts plays an important role for information
analytics in big data. This paper surveys research work on
text-based event temporal resolution and reasoning including
the identification of events, temporal information resolutions of
events, the rule-based temporal relation reasoning between events
and the temporal representations. We point out the shortcomings
of existing research work and the future trends for advancing the
identification of events and the establishing/reasoning of temporal
relations in the future.
Index Terms—Event; Information Analytics; Temporal Infor-
mation; Resolution; Reasoning; Big Data
I. INTRODUCTION
With the development of information technology, we have
entered the big data era. Very large-scaled data in various
types such as text, image, video and audio are generated and
collected continuously in Internet of Things. Believe it or not,
big data and related technologies have greatly changed and will
continue to affect our life dramatically, and it becomes a great
challenge to process and analysis the big data for intelligence.
Key technologies for big data processing include informa-
tion semantic organization and information analysis for insight
and decision making. It is difficult to organize multi-sourced
and heterogeneous information semantically for satisfying var-
ious information needs. Traditional relational databases have
met some problems for the management of multi-sourced and
heterogeneous information, because the schema is complex
and dynamic. The resource types and their relations may
evolve with time, and it is impossible to modify the schema
in the relational database frequently because the tight relations
between the databases and applications. One possible way is
using hierarchical model (e.g., tree model) and network model
to organize the multi-sourced and heterogeneous information.
So novel semantic data model to satisfy the characteristics of
multi-sourced and heterogeneous information in the big data
era [1], [2].
Events happen every day and every time in the world.
Usually, an event contains the following elements: when,
where, who, what, why and how. When and where are the
basic information which represents the time and location of
an event; who and what describe the content of an event; and,
why and how are optional elements for describing the detail
of an event.
Event could be used as a semantic unit to organize infor-
mation based on the semantic relevance between information.
Information of related things in an event could be organized
together, which is naturally suitable for archive and search
because people memorize the things by attaching them to
the corresponding events. During the event-based information
organization, time is a naturally sequential clue for organizing
information in the context of the event.
Event still lacks a unified definition. For example, in linguis-
tic research, an event could be a verb or a noun. In industrial
control, an event may be a change of status. Generally, an
event means that when, where, who do what, how and why.
Although there are different definitions of events, all the events
are unavoidably relevant to two essential elements or attributes
— time and location. It is ubiquitous to query and analyze the
temporal and geographic information of events. Furthermore,
time is an essential dimension for any information space.
Because time has naturally sequential characteristic, the
temporal attributes of events could be used to organize infor-
mation based on events in sequence. The temporal sequence
of events could be used to browse, search and answer time
related questions. Therefore, it is important to monitor and
trace events, and the temporal information is necessary to
formulate the sequence of events.
In this paper, we focus on temporal information resolution
and temporal relation reasoning among events implied in
texts. Massive information and knowledge are represented by
natural languages and applied in texts. During the information
processing, images, audios and videos are also annotated
by natural languages. So the text-based event identification,
extraction and reasoning will be important for information
analytics based on big data.
II. I
DENTIFICATIONS OF EVENTS AND TEMPORAL
INFORMATION
Research on event and temporal analysis in the text is
distributed in multidisciplinary.
• In philosophy study, temporal analysis research work
mostly focuses on event, temporal information and tem-
poral reasoning between two events, between an event
and a temporal representation, or between two temporal
representations [3]–[6].
• In linguistics study, research work mainly focuses on the
temporal sequence of a single event involving temporal
2015 International Conference on Identification, Information, and Knowledge in the Internet of Things
978-1-4673-8637-1/15 $31.00 © 2015 IEEE
DOI 10.1109/IIKI.2015.24
78