A Typology of Temporal Data Imperfection

Nassira Achich

, Fatma Ghorbel

1,2

, Fayc¸al Hamdi

, Elisabeth Metais

and Faiez Gargouri

MIRACL Laboratory, University of Sfax, Tunisia

CEDRIC Laboratory, Conservatoire National des Arts et M

etiers (CNAM), France

Keywords: Data Imperfections, Typology of Temporal Data Imperfection, Direct and Indirect Imperfections, Natural

Language based Temporal Data, Context - Dependent Temporal Data.

Abstract:

Temporal data may be subject to several types of imperfection (e.g., uncertainty, imprecision..). In this context,

several typologies of data imperfections have been already proposed. However, these typologies cannot be

applied to temporal data because of the complexity of this type of data and the speciﬁcity that it contains.

Besides, to the best of our knowledge, there is no typology of temporal data imperfections. In this paper, we

propose a typology of temporal data imperfections. Our typology is divided into direct imperfections of both

numeric temporal data and natural language based temporal data, indirect imperfections that can be deduced

from the direct ones and granularity (i.e., context - dependent temporal data) which is related to several factors

that can interfer in specifying the imperfection type such as person’s proﬁle and multiculturalism. We ﬁnish

by representing an example of imprecise temporal data in PersonLink ontology.

1 INTRODUCTION

Temporal data is an important aspect in representing

ontologies such as PersonLink (Herradi et al., 2015)

which is a multicultural and multilingual OWL 2 on-

tology for storing, modeling and reasoning on family

relations. However, temporal data are subject to var-

ious kinds of imperfection. For instance, we ﬁnd ”I

lived in Berlin before changing my job”, ”I traveled

to Spain during 2005” and ”My sister married before

moving to Paris”. Several typologies of data imper-

fections have been proposed. Some are generic (e.g.,

typology proposed by (Bouchon-Meunier, 1993) )

and others are speciﬁc to some domains (e.g., typol-

ogy of imperfections adapted to the context of archae-

ological data proposed by (Desjardin et al., 2012) ).

However, to the best of our knowledge, there is no ty-

pology of temporal data imperfections. Temporal data

can have more imperfections compared to the ones

proposed in the generic typologies. Thus, these lat-

ters are inadequate to be adapted to temporal data.

In this paper, we propose a typology of tempo-

ral data imperfections. It is classiﬁed into direct im-

perfections of both numeric temporal data and natu-

ral language based temporal data, indirect imperfec-

tions that can be deduced from the direct ones and

granularity (i.e., context - dependent temporal data)

which is related to several factors such as person’s

proﬁle and multiculturalism. The direct imperfections

can be deduced directly from the given data: uncer-

tainty, typing error, imprecision, missing and useless-

ness. The Indirect imperfections are those that can be

generated from the direct ones. We distinguish three

types of indirect imperfections: (i) The incoherence

can be generated from the uncertainty and typing er-

ror. (ii) The incompleteness can be generated from

the imprecision and the missing. (iii) The redundancy

can be generated from the uselessness. Granularity is

the general context of the given temporal data which

can determine the type of imperfection.

The rest of the paper is organized as follows. Re-

lated work is reviewed in the next section. Then, we

introduce our proposed typology of temporal data im-

perfections in Section 3. In Section 4, we represent

an example of imprecise temporal data in PersonLink

ontology. Section 5 summarizes our work and gives

some future directions.

2 RELATED WORK:

TYPOLOGIES OF DATA

IMPERFECTIONS

Related work about the typologies of data imperfec-

tions is discussed in this section. We distinguish two

Achich, N., Ghorbel, F., Hamdi, F., Metais, E. and Gargouri, F.

A Typology of Temporal Data Imperfection.

DOI: 10.5220/0008168403050311

In Proceedings of the 11th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2019), pages 305-311

ISBN: 978-989-758-382-7

305

types of typologies: generic typologies of data imper-

fections and speciﬁc typologies of data imperfections

in a given domain.

2.1 Generic Typologies of Data

Imperfections

We identify four generic typologies of data imperfec-

tions:

• (Niskanen, 1989) proposes a typology of non-

precision of the data. He deﬁnes four concepts

which are uncertainty, imprecision, ambiguity and

generality. Uncertainty is a concept associated

with the accuracy and error deﬁnitions. Impre-

cision exists when objects are naturally imprecise

or the verbal expression can have several mean-

ings or when a human cannot identify an object

exactly. Ambiguity appears when there are sev-

eral points of view about the same subject of the

real world. Generality is the multiple representa-

tion of reality according to the level of detail.

• (Bouchon-Meunier, 1993) in her ﬁrst work dis-

tinguishes two types of imperfection which are

uncertainty and imprecision. In her second work

(Bouchon-Meunier, 1995), she distinguishes a

third factor of imperfection, which is the incom-

pleteness. She deﬁnes uncertainty by the validity

of the data. The imprecision of the data is due to

the vague or approximate nature of the used se-

mantic. Incompleteness is related to incomplete

data or lack of data.

• (Klir and Yuan, 1995) propose a typology of

data uncertainty. The authors divide the uncer-

tainty into two classes which are fuzziness and

ambiguity. Ambiguity refers to conﬂict and non-

speciﬁcity.

• (Smets, 1997) establishes a classiﬁcation of the

imperfection of the data. It is divided into three

groups which are imprecision, inconsistency and

uncertainty. The imprecision is relative to the

content of the data that may or may not have er-

rors. Inconsistency is related to conﬂicting. Un-

certainty can be subjective (i.e. observer’s opin-

ion) or objective (i.e. data properties).

2.2 Speciﬁc Typologies of Data

Imperfections

We identify seven typologies of data imperfections

speciﬁc to a given domain.

• (Gershon, 1998) focuses on kinds of imperfec-

tion in information that might be provided to an

analyst or decision maker. He proposes a taxon-

omy of causes for imperfect knowledge of the in-

formation state. The ﬁrst level of the taxonomy

contains six inputs which are incomplete info,

inconsistency, info too complicated, uncertainty,

corrupt data/info and the quality of the presenta-

tion.

• (Fisher, 2005) propose a typology of uncertainty

of geographic data. They classify the data into a

well or a badly deﬁned data. In the case where the

data is well deﬁned, it is subject to uncertainty. In

other cases, the data is poorly deﬁned and the im-

perfection of the data is due to imprecision, ambi-

guity and / or incompleteness. He focuses on am-

biguity. It occurs when there is a doubt about how

to deﬁne an object or phenomenon. Two types of

ambiguities are recognized namely disagreement

and lack of speciﬁcity.

• (Olteanu, 2006) propose a typology based on

(Fisher, 2003) and (Fisher, 2005) to classify the

imperfection of a set of textual data describing

ethnographic objects. Four types of imperfec-

tions are distinguished: Imprecision, uncertainty,

level of detail and incompleteness. Imprecision

concerns the difﬁculty of expressing knowledge

clearly and precisely. The uncertainty concerns a

doubt about the validity of information. The level

of detail is knowledge presented in several gran-

ularities. Incompleteness refers to the absence of

information.

• (Casta, 2009) establishes a typology of the imper-

fection of the data resulting from the economic

activity. It is divided into uncertainty, impreci-

sion and error. Uncertainty refers to the state of an

agent who does not know about the future because

the set of possible events, or contingencies, in-

cludes more than one element and these elements

do not obey strict determinism. Imprecision refers

to a lack of rigor or an operational constraint that

affects measurement or, most often in the social

sciences, an ambiguity inherent in the formulation

of concepts. The error is deﬁned as the dispersion

of probable gains and losses around their mean.

• (Desjardin et al., 2012) rely on the typology of

(Fisher et al., 2005) to propose a typology of im-

perfection adapted to the context of archaeolog-

ical data. They classify imperfections into un-

certainty, imprecision, ambiguity and incomplete-

ness. The uncertainty occurs when there is a

doubt about the validity of knowledge. Impre-

cision is the difﬁculty in expressing knowledge

clearly. The ambiguity is the difﬁculty in agree-

ing. Incompleteness is the fact that there is miss-

KEOD 2019 - 11th International Conference on Knowledge Engineering and Ontology Development

306

ing or partial knowledge.

• (M. Snoussi, 2014) propose a classiﬁcation that

concerns a large set of imperfections that was built

speciﬁcally to characterize spatial data. This tax-

onomy distinguishes three types of imperfection:

Imprecision occurs when the true value is located

in a deﬁned subset of values, inconsistency is the

conﬂict or incoherence in the values and uncer-

tainty is the partial knowledge about the true value

of information.

• (Sta, 2016) proposes several types of imperfect

data during the process of information retrieval

and data integration in smart cities. This im-

perfection can have several forms: Uncertain in-

formation reﬂects the lack of knowledge, impre-

cision information translates the non-speciﬁcity,

vague information reﬂects an ambiguous infor-

mation and missing information reﬂects the not

found or incomplete information.

2.3 Discussion

There is a big number of typologies of data im-

perfection. Authors such as (Niskanen, 1989) and

(Bouchon-Meunier, 1995) start by proposing generic

typologies. Then, to respond to several domains, au-

thors such as (Fisher, 2005), (Desjardin et al., 2012)

and (Sta, 2016) propose typologies for speciﬁc ﬁelds.

However, some correspond better to a reality than

others. Also, there is not a deﬁnitive terminology

and deﬁnition of terms used to qualify imperfect data,

such as uncertainty and imprecision. We also note

that the majority of these typologies share three com-

mon concepts which are the imprecision, the uncer-

tainty and the incompleteness. Imperfection types

are interdependent. According to (Bouchon-Meunier,

1995), ”Incompleteness leads to uncertainties, im-

precision can also be associated with incomplete-

ness and they generate uncertainties during their han-

dling”. According to (Smets, 1998), imprecision al-

ways refers to incompleteness. Imprecision may be

the source of uncertainty, but not necessarily (Smets,

1998).

Temporal data can have more imperfections com-

pared to the ones proposed in the generic typologies

because of the complexity of this type of data and the

speciﬁcity that it contains (i.e., It can be numeric or

natural language-based. It can also depend on the

general context of the information that contains the

temporal data and they may also be subject to several

factors such as the multiculturalism). Thus, generic

typologies cannot be adapted to temporal data (i.e,

they are inadequate). For instance, if we have the in-

formation ”I forget when last time I visited my uncle

who lives in Japan ”, the temporal data indicates a

”Missing” which is a type of imperfection that no one

of the generic typologies contains. Another example

is ”The ﬁrst day of the week, we will have a meeting”.

In this example the temporal data indicates a ”circum-

locution” which is another type of imperfection that

no one of the generic typologies contains. Finally, to

the best of our knowledge, there is no typologies ded-

icated to temporal data imperfection.

3 OUR TYPOLOGY OF

TEMPORAL DATA

IMPERFECTION

In this section, we introduce our typology of tempo-

ral data imperfections illustrated by Figure 1 . Our

typology is based on the studied mentioned typolo-

gies and collected real examples. We divide our ty-

pology into direct imperfections, indirect imperfec-

tions and granularity. The direct ones are those which

can be deduced directly from the given data: uncer-

tainty, typing error, imprecision, missing and useless-

ness. The Indirect imperfections are those that can be

deduced from the direct ones. We distinguish three

types of indirect imperfections: (i) The incoherence

can be generated from the uncertainty and typing er-

ror. (ii) The incompleteness can be generated from

the imprecision and the missing. (iii) The redundancy

can be generated from the uselessness. Granularity is

the general context of the given temporal data which

can determine the type of imperfection.

3.1 Direct Imperfection Types

We distinguish seven direct imperfection types: un-

certainty, typing error, imprecision, missing, circum-

locution and uselessness which will be detailed one

by one and illustrated by examples.

• Uncertainty. According to (Smets, 1999), each

statement is whether true or false. However, the

knowledge about the statement, if it is uncertain,

does not allow you to decide either it is true or

false. Uncertainty is partial knowledge of the true

value of the data. It is the lack of information.

In our typology, uncertainty has two categories.

The ﬁrst category includes uncertain data that re-

spect the rules of common sense. It can be due

to a doubt. As an example, ”I’m not sure if it

was the last summer or the one before when my

aunt came from Paris to spend the holidays with

us”. In this example, the person is doubting about

the given temporal data. However, the latter re-

A Typology of Temporal Data Imperfection

307

Figure 1: Our typology of temporal data imperfection.

spects the rules of common sense. The second

category includes unreliable uncertain data in re-

lation to the general rules of common sense. For

example, ”We swam in the sea during the new

year holiday”. This data does not respect the rules

of common sense since the new year holidays are

in winter and not in summer.

• Typing Error. In some cases like using a sys-

tem, the entered data by the user may be subject

to errors. For instance, if we have the informa-

tion ”The municipality employee typed my date

of birth wrongly”, the inserted date of birth is a

typing error imperfection.

• Imprecision. Most of the temporal data are im-

precise. Imprecision leads to the ambiguity about

the information. Imprecise temporal data has two

categories. The ﬁrst one includes the ones that

are translatable to perfect data. For example, ”We

moved to Tunis by 2008”.”by 2008” is a numeric

temporal data that can be translatable to a perfect

interval that has two bounds (2007 .. 2009). An-

other example, if we have the information ” Dur-

ing the summer vacation, they travelled to Spain

”. ”During the summer vacation” presents a nat-

ural language-based temporal data which can be

translatable to a perfect temporal interval [01/06

.. 31/08] since the bounds are precise. The

second category includes the ones that are non-

translatable to perfect data. Several examples can

be introduced, such as ”I moved to my new job

after my daughter’s birth”; ”After my daughter’s

birth” is imprecise since her daughter’s birth is un-

known. Another example is ”He was sick before

his death”; ”Before his death” is also imprecise

since we do not know how much time before his

death he was sick. Another type of natural lan-

guage based temporal data that is non-translatable

to perfect data is the adverbs of time such as ”

Nowadays”, ”Currently”, ”Now” and ”later”.

• Missing. Temporal data may be subject to a

missing. Missing can be either partial or total. For

example, the following data ”It was in March but

I forget the year”. In this example, the missing

is partial. The person remembers the month how-

ever he forgets the year. Another example, if we

have the information ”I forget when last time I vis-

ited my aunt who lives in Paris”. In this example,

the missing is total. The person forgets totally the

temporal data.

• Circumlocution. Temporal data can be the ob-

ject of a circumlocution. For example, ” The ﬁrst

day of the week, we will have a work meeting”.

The person used ” the ﬁrst day of the week” state-

ment, however he means ”Monday”. Another ex-

KEOD 2019 - 11th International Conference on Knowledge Engineering and Ontology Development

308

ample, ”The marriage is two days after the week-

end”. The person replaces ”Tuesday” by the state-

ment ”two days after the weekend”.

• Uselessness. This imperfection means that the

given information is useless and there is not any

added value to the given information. For ex-

ample, if we have the information ”During the

Christmas in December, people celebrate all over

the world”, it is very enough to say ”During the

Christmas” since it is a general truth that the new

year is obviously in December. Therefore, in this

case, ”in December” presents a useless temporal

information.

3.2 Indirect Imperfection Types

We distinguish three indirect imperfection types: in-

coherence, incompleteness and redundancy. These

imperfections are deduced from the direct imperfec-

tion types.

• Incoherence. To deﬁne the incoherence, we are

based on on (Gavignet et al., 2016). According

to her, the incoherence refers to ”the existence of

contradictory data on the same object”. Thus, un-

certainty and typing error that can exist in a given

temporal data may lead to the incoherence.

• Incompleteness. Missing and imprecise tempo-

ral data may lead to an incompleteness. For in-

stance, if we have the information ”My uncle will

come next weekend”; ”next weekend” is impre-

cise and, so incomplete. It could be ”Saturday” or

”Sunday”.

• Redundancy. It can be concluded when use-

lessness interferes in the given temporal data. For

instance, if we have the information ”A long time

ago, my grand-parents dead. It was several years

ago”, in this statement the person indicates two

temporal data, ”A long time ago” and ”several

years ago”, that give the same meanings, which

is useless and redundant.

3.3 Context - Dependent Temporal Data

Granularity presents the scale or level of detail in a

set of data. The context of a given temporal data de-

termines the type of imperfection. For instance, if we

have ”We can meet next weekend”; ”next weekend”

is imprecise. It could be ”Saturday” or ”Sunday” or

both. However, if we have ”We will spend next week-

end in Berlin”, it is clear from the context that “next

weekend” means the whole weekend (i.e., both of Sat-

urday and Sunday) and thus the data is precise. Be-

sides, several factors may interfere in determining the

type of imperfection such as the proﬁle of person and

multiculturalism.

• The Proﬁle of the Person. From the person

themselves we can decide the type of imperfec-

tion. For instance, the temporal data given by an

Alzheimer’s patient is probably uncertain. For ex-

ample ” I am not sure it was in the morning or at

night ”. Another example concerns the children

who frequently get confused about ”yesterday”,

”tomorrow” and ”today”.

• Multiculturalism. Multiculturalism aspect re-

ﬂects the cultural differences where ideas are fo-

cused on the beliefs ways of societies. For in-

stance, ”the weekend” in Europe is ”Saturday and

Sunday”. It differs from the one in the Arab World

which is ”Friday and Saturday”.

4 REPRESENTING EXAMPLES

OF IMPRECISE TEMPORAL

DATA IN PERSONLINK

ONTOLOGY

In this section, we represent some examples of im-

precise temporal data in PersonLink

ontology based

on our previous work (Ghorbel et al., 2018) in which

we extend the 4D-ﬂuents approach (Welty, 2006) to

represent imprecise time intervals and crisp temporal

interval relations.

Let’s have the following example: ”Peter was

married to Stephanie just after he was graduated with

a PhD and it lasts 15 years. Peter was graduated with

a PhD in 1960”. This example contains a natural

language-based temporal data which is ”just after he

was graduated” . It is an imprecise temporal data. Let

I = [I

−

, I

] be an imprecise time interval representing

the duration of the marriage of Peter and Stephanie.

Assume that I

−

= {1960...1963} presents the inter-

val of the beginning bound and I

= {1975...1978}

the interval of the ending bound. Figure 2 illustrates

a part of PersonLink ontology which represents this

example.

Another example, we consider the following in-

formation: “John was married to B

eatrice since about

10 years. John was married to Maria just after he

was graduated with a PhD and it lasts 15 years. John

was graduated with a PhD in 1980”. Let I = [I

−

, I

]

and J = [J

−

, J

] be two imprecise time intervals rep-

resenting, respectively, the duration of the marriage

of John and B

eatrice and the one of John and Maria.

http://cedric.cnam.fr/ hamdif/ontologies/ﬁles/

PersonLink.html

A Typology of Temporal Data Imperfection

309

Figure 2: Representation of the ﬁrst example of imprecise

temporal data in PersonLink.

Assume that I

−

= {2007...2009}, I

= 2018, J

−

{1980...1983} and J

= {1995...1998}. Figure 3 il-

lustrates this example.

Figure 3: Representation of the second example example of

imprecise temporal data in PersonLink.

5 CONCLUSION AND FUTURE

WORK

In this paper, we introduce our typology of tempo-

ral data imperfection. It is classiﬁed into direct im-

perfections of both numeric temporal data and natural

language based temporal data, indirect imperfections

that can be deduced from the direct ones and granular-

ity (i.e., context - dependent temporal data) which is

related to several factors such as person’s proﬁle and

multiculturalism.

Direct imperfections consist of uncertainty, typing

error, imprecision, missing, circumlocution and use-

lessness. Indirect imperfections are generated from

the direct ones. Uncertainty and typing error lead to

the Incoherence. Imprecision and missing lead to in-

completeness. Uselessness leads to redundancy.

We ﬁnish by representing an example of imprecise

temporal data in PersonLink ontology.

As for continuity of our work, we plan to pursue

on treating the other proposed types of temporal data

imperfections in our typology starting by the uncer-

tainty.

REFERENCES

Bouchon-Meunier, B. (1993). La logique ﬂoue.

Bouchon-Meunier, B. (1995). La logique ﬂoue et ses appli-

cations.

Casta, J. F. (2009). Incertitude et comptabilit

Desjardin, E., Nocent, O., and De Runz, C. (2012). Prise

en compte de l’imperfection des connaissances depuis

la saisie des donn

ees jusqu’

a la restitution 3d. es-

tablished by: Mauro Cristofani and Riccardo Fran-

covich, (supplemento 3):385–396.

Fisher, P., C. A. . R. R. W. (2005). Nature de l’incertitude

pour les donnees spatiales. pages 49–64. Qualit

de l’information g

eographique, Paris : Herm

es et

Lavoisier.

Fisher, P. (2003). Models of uncertainty in spatial data.

pages 191–203. Geographical information systems.

Gavignet,

E., Leclercq,

E., Cullot, N., and Savonnet, M.

(2016). Raisonner en logique modale sur l’incertitude

ee aux donn

ees-application en arch

eologie. Revue

Internationale de G

eomatique, 26(4):467–490.

Gershon, N. (1998). Visualization of an imperfect world.

pages 43–45. IEEE Computer Graphics and Applica-

tions, 18(4).

Ghorbel, F., M

etais, E., and Hamdi, F. (2018). A crisp-

based approach for representing and reasoning on im-

precise time intervals in owl 2. In International Con-

ference on Intelligent Systems Design and Applica-

tions, pages 640–649. Springer.

Herradi, N., Hamdi, F., M

etais, E., Ghorbel, F., and

Soukane, A. (2015). Personlink: an ontology rep-

resenting family relationships for the captain memo

memory prosthesis. In International Conference on

Conceptual Modeling, pages 3–13. Springer.

Klir, G. J. and Yuan, B. (1995). Fuzzy sets and fuzzy logic:

theory and applications, volume 574. Prentice Hall

PTR New Jersey.

M. Snoussi, P.-A. D. (2014). Methodological proposals to

handle imperfect spatial and temporal information in

the context of natural hazard studies. pages 495–517.

International Journal of Geomatics and Spatial Anal-

ysis 3-4, 23.

KEOD 2019 - 11th International Conference on Knowledge Engineering and Ontology Development

310

Niskanen, V. A. (1989). Introduction to imprecise reason-

ing, uncertainty, decision making and knowledge en-

gineering.

Olteanu, A. M., M. S. R. A. . D. E. (2006). L’apport de

donn

ees spatiales pour une base de donn

ees ethno-

graphique. Actes de Conf

ence SAGEO.

Smets, P. (1997). Imperfect information: Imprecision-

uncertainty en uncertainty management in informa-

tion systems: from needs to solutions. pages 225–254.

Smets, P. (1999). Imperfect information: Imprecision-

uncertainty en uncertainty management in informa-

tion systems: from needs to solutions. amuhai motro

and phillip smets.

Sta, H. (2016). Quality and the efﬁciency of data in “smart-

cities. Future Generation Computer Systems.

Welty, C., . F. R. (2006). A reusable ontology for ﬂuents

in owl. pages 226––336. 4th Int’l Conf. on Formal

Ontology in Information Systems (FOIS 2006).

A Typology of Temporal Data Imperfection

311