KABBALAH: ON SPATIO-TEMPORAL DATABASE VISUALIZATION WITH HISTORICAL EVENTS: A CASE STUDY OF HISTORY FLOW OF CHIA-YI BEIMEN STATION

This system is made for researchers who study pattern of city or spatial transformation by using computational way to interpret data logically. In order to make use of all historical data with GIS in system, an exact metadata is necessary and needed to build first. The Cubism project is aimed to presume how different historical data normalized to become information in spatio-temporal database. To make temporal map have higher capability with presentation of history context.


INTRODUCTION 1.1 Motivation
A city formed with numerous classes of historical events.However these records can not completely describe a timespace subject of one city.There are some methods to present historical events by combining with timeline or GIS to present its order or distribution position.But to sense and understand one city only according to single dimension of time or spatial is not objective enough.
A historical research depends on reading lots of historical data to know well with the target which research focus on.This progress is not only time consuming for researchers who reading history, but also can not avoid to collect wrong information.
With the same reading behavior, reading a hypertext is more strength than a linear one.(Chen, 2000).For instance, a GIS map accompanied with occurrence order to translate events information into hypertext or even hypermedia.If the traditional behavior combines with these computational methods, to understanding the historical background of one space will be more efficient.

Purpose
The main purpose of this study is using a symbol system to represent the time-space subject.A conceptual and iconify system may be applied to comprehend the duration of development process.A specific time-space flow will be viewed when focus one target or set rules for seeing complete picture of this city's history.Here are the two goals of this system: 1.To get the biography of one target: Different events may have relationship with each other.By using texts can not present the context of history clearly.
2. To compare individual with others: Multiple dimensions are hard to interpret or make a contrast.The individuals must differentiate before comparing with each other.

Time Geography
"Time and space" is an indivisible unity.In social culture domain, a space issue should be discussed with history.Time is generated by interaction of society in one space.What people see the development of society is the objective definition for time and space.
During a period of time, space or buildings may change by events.These changes are considered as the attributes of timespace subject.If these changes didn't influence the existence of time-space subject, the new subject may treat as the continued subject for this space.This calls material continuity.
Events may include changed attributes of subject with a series orders.Time is able to measure these orders by defining measurement, like known units such as year, month, day, hours, minute, and second.(Robert Levine, 1997) Events or facts occurred in past are hard to trace back.Even the most advanced technology is not able to record all attributes of all objects in a city at the same time.

Temporal Database
For a systematic way to record events and changes, Space-Time Composites Model (STC) used to ameliorate the snapshot record model.STC model separates components are about to change, as Figure 1 shows, the relationship between rural and urban.The basic units of this model are the components divide by changes.This model need only to save the value of attributes for these changed components and may solve the consuming problem with working hours and resources.The specific component may be helpful for tracing the different units in not the same time.However, this way still has a potential problem.
All the components have to redefine when one space changed.This model is not suitable for constantly inputting data.(Langran & Chrisman, 1988) Figure 1.Observing changing between rural and urban in different time (from Langran, Chrisman, 1988, redraw by this study) Peuquet and Duan (1995) proposed Event-based Spatiotemporal Data Model (ESTDM) for processing temporal information.By storing events as a unit of temporal data, these information can be integrated with spatial information, and unfold on map.
Leslie mentioned about one temporal data would descript as temporal information (when), spatial information (where), and non spatio-temporal information (what changed).(Leslie, C., Barnes, G., Binford, M., Smith, S., 2001) The contribution of temporal database for history research includes: (1) To understand the rules that govern changes observed in the world; (2) To explain the state of the world at some earlier time; (3) To predict the state of the world in the future; and (4) To plan a sequence of actions that will lead to a desirable future.
These four points make urban development research be able to observe the future trends form evolvement, and adjust the plan or strategy of one city.

METHODOLOGY
The memories of one city's development are hided in historical buildings in this city.People may use these historical buildings to be junctions connect past and present (as Figure 2).As historical building play a role of objective subject, this way makes observing the development of entire city by examining every historical event that this historical building experienced see the changing of space in multiple aspects.
1. Expound existence and essence of temporal map. 2. Plan metadata and normalization of spatio-temporal database

PRIMARY RESULT
The cue-point represents something changed, but it not means all this building or space make a reconstruction.Evan a small part repaired or the function of this building had been changed, these are able to be called "variation."A symbolic represent system may help to present different variation.
By using structure of timeline, every single attribute has its subtimeline.As any attribute of a subject changed, one of a subtimeline shows a cue-point, and the main timeline may present the same particular time.For one subject, its state is stationary between two cue-points of timeline in different time.Every change of an attribute of one subject should mark a cue-point, and also there are multiple attributes changed probably.As Figure 3 shows below: Type class includes event type, action, and complement.Events type are classified to "social dynamics"(SD) which people cause environment they live changed, "disaster damage" (DD) which natural phenomena makes object transform, "territorial features"(TF) which are initial conditions of space for defining the original state in the beginning of timeline, "urban policies"(UP) which are supervisors or strategies make city change.Complement is supplement for describing which not mentioned from above metadata.
In order to trace not only current value, but also changes in multiple cue-points, attribute class notes changed item, and the value changed before and after.If there are multiple attributes changed in one event, these data should log in multiple and sequentially way also for recording totally.The value records are helpful for observing if there is any other event not logged.For instance, if there are two cue-points of timeline, and the new (after) value of the first cue-point is not the same as the original (before) value of the second point, this proves some events happened between these two cue-points and make things change.The values in different time can inspect the processing of one subject or one object in period of time.The record class notes established time and establisher, or modified information of this data.
These events data can be connected with the target descriptions, and generated historic events chain.The historic events chain combined with vertex (event) and edge (relationship).Like directed graph (or Digraph) in graph theory, every relationship has its "cause" and "effect" with sequential.The after events are not able to affect the previous ones.In other words, the start node and end node of one edge will not be the same.The "loop" will not possible to appear in one vertex, and "cause" can not be "effect" at the same time.For these rules, the historic events chain will present as a simple graph.
There is relevance between the four event types mentioned above.One event could be the cause of any other events, but kinds of event types can not give rise to other kinds of event type.There are initial rules for these events combining into chains.Like urban policies and social dynamics type could not cause a disaster damage type event.On the contrary, a disaster damage type event may bring about the policies or social changes.However, social dynamics type event will be influenced by other three types.(as Figure 4) The events related to Beimen station list in The cue-point of one timeline means change appear on an object.There will be two state generate by an event.An event presents year and action with rectangle as Figure 5 shows below.
States shows name of target (might be initiators or recipients, called pronoun for an instance (as Figure 6).The final result with all events related to Beimen shows as Figure 7 below:

CONCOLUSION
Every existing object, like mankind, area, or buildings all have their own timeline.So do a city space.These objects in same city place will generate relationships as time goes by.The historic context analysis research has computational way to observe the subjective perspective of different object.This is the main purpose of this study.
Establishing this historic event database not only supports the historic space research, but also another way to make preservation for culture heritage in digital method.If there is a good interface and right way to use this database, this system may become a local historic education system to pass the right historic context knowledge on, and to make culture of one city flourish.
Figure 2. Historic building used as the interface from past to current

Figure 3 .
Figure 3.The state of object and the relationship between object and subject (from Langran, G., Chrisman, N., R., 1988, redraw by this study)

Figure 5 .
Figure 5.The event present in symbol with year and action information ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume II-5/W3, 2015 25th International CIPA Symposium 2015, 31 August -04 September 2015, Taipei, Taiwan Table 1 as below:Figure 4. The influence relationship between four types of events