Assertion

The following subsection takes up the theory of temporal and bitemporal data storage on page 80 and focuses on the investigation of the claimed correct version of data and statements, assertion and context with the management of time. The term assertion time refers to the length of time in which lines of a database table or the creator of this data claim that this version of the data is the true and most current. As a rule, it is the time span in which the data is active and visible in the real world. As soon as these lines are no longer visible in the real world, they also lose the right to be the true version (Johnston, 2014, p. 31). The change takes place by modifying these active lines. This can be done in three ways by deleting rows or data: Set a deletion flag, Physical Deletion, or Logical Deletion. If data in a table is flagged for deletion, it is not physically deleted, but is considered inactive and no longer true from the start of the deletion flag. WithPhysical Deletion, also calledHard Deletion(Ghosh, 2013), the row is completely removed from the table, thus simultaneously erasing the meaning of the data and the assertion. The third variantLogical Deletion, also calledSoft Delete(Ghosh, 2013), deletes neither the row nor its meaning, but its assertion is deleted (Johnston, 2014, p. 31). Summarized and based on the theory of bitemporal data it can be stated that data is valid and in this case also veritable if it appears actively in a table or in the real world and loses this status when it is deleted. Since this time span is determined by the change times, insertion and deletion of the data, their designation as transaction time would be obvious.

Johnston (2014, p. 225) is of the opinion, however, that assertion time represents a more precise description of this term and is therefore the more suitable name for it (Johnston, 2014, p. 225).

Inserted rows in data tables are each stored in the system with an assertion time from the time of insertion, Now(!), into the bitemporal table until a fixed time in the future, 9999. As soon as this row is deleted, the end date changes to the time of removal of the row, Now(!).

Information is therefore only the true version in a certain period. Johnston (2014, p. 226) shows with the so­calledThe Six­Fold WayHow six different views of information and state­

ments are possible.

These views stand in combination with the three time forms, past, present and future, and on the one hand with a past assertion of the viewer and on the other hand with the present

5.1 Design building process of the model | 83 assertion of the viewer (see table 5.1). The first line of the table contains statements about things that already lie in the past. The second line represents assertions about things that are still relevant. The last line shows statements about things that are still to come. Johnston wants to show that information in six versions can claim to be the true version (Johnston, 2014, p. 225­226).

Table 5.1 the six­fold way (ad. Johnston, 2014, p. 226)

what we used to assert what we currently assert

what things used to be like

(1) what we used to assert things used to be like.

(4) what we currently assert things used to be like.

what things are like now (2) what we used to assert things are like now.

(5) what we currently assert things are like now.

what things will be like (3) what we used to assert things will be like.

(6) what we currently assert things will be like.

Once the assertion time of a datarow has expired, the information it contains is no longer the valid version of that information. Thus, before the assertion time, the information is not yet the valid version, but becomes so after the specified time. Johnston (2014, p. 230) extended this approach. It criticizes that only statements of the past and the present are con­

sidered. Future information about the condition of an object is contained in the previous concept, but not future claims. Accordingly, he has extended the previous model by a col­

umn that takes this missing component into account. Due to the three possibilities about the state of things to a period and the three possibilities about claims to things to a period there are now altogether nine combinations respectively nine views on information. This extended approach is calledThe Nine­Fold Way(Johnston, 2014, p. 230).

Table 5.2 The nine­fold way (ad. Johnston, 2014, p. 230)

what we used to assert what we currently as­

sert

what we will assert

what things used to be like

(1) what we used to assert things used to be like.

(4) what we currently as­

sert things used to be like.

(7) what we will assert things used to be like.

what things are like now

(2) what we used to assert things are like now.

(5) what we currently as­

sert things are like now.

(8) what we will assert things are like now.

what things will be like

(3) what we used to assert things will be like.

(6) what we currently as­

sert things will be like.

(9) what we will assert things will be like.

The table 5.2 shows that data over time can travel through different time periods. Data that is compared with each other must therefore be of the same time type to analyze versions of data without any doubt. The versioning of data is therefore of great importance and must be distinguished from historization. While historization means the storage of data for his­

torical analyses, versioning means that the claimed version is determined beyond doubt at a certain point in time. This version was only presumed in the case of the observed use cases in practice. For the versioning Johnston (2014) suggests anAsserted Versioning Frameworkin which the bi­temporal time collection with two timestamps is extended by two more times.

The assertionstartandendtimestamps. Looking at the proposed principles, it becomes clear that the management of time in the context of data management can be considered of great importance. In the category ofcoherencethese are

• time,

• traceability,

with the categorycontents

• actuality,

• universality,

• integrity format,

• comprehensibility,

and finally with the categoryfunction

• versioning,

• historicization,

• measurability,

• processability.

It becomes clear that the time as mentioned as a single principle in the field of coherence contains much more than just the coherence category itself. From this fact a variety of data quality problems can be assumed in the management of time series to data. If data are con­

densed into information with each other, this can only be done comprehensibly if the temporal states are the same. To do this correctly, according to the consideration of Johnston (2014), four time stamps would have to become necessary. With all investigations for the abstraction of the problem with the preliminary studies in addition, in the discussion with interviewees it was never confirmed that a versioning of the data takes place with such a time management.

It is assumed that this circumstance is not yet considered in the existing approaches of the data definition.