Hana Alouaoui, Sami Yassine Turki
LTSIRS Laboratory of Remote Sensing and Information Systems with Spatial Reference
ENIT National Engineering School of Tunis, Tunis El Manar University, Tunis, Tunisia
Sami Faiz
LTSIRS Laboratory of Remote Sensing and Information Systems with Spatial Reference
ENIT National Engineering School of Tunis, Tunis El Manar University, Tunis, Tunisia
Keywords: Spatiotemporal database, Topological relationships, Spatiotemporal data mining, Data mining query
Abstract: This paper presents an approach for mining spatiotemporal association rules. The proposed method is based
on the computation of neighborhood relationships between geographic objects during a time interval. This
kind of information is extracted from spatiotemporal database by the means of special mining queries
enriched by time management parameters. The resulting spatiotemporal predicates are then processed by
classical data mining tools in order to generate spatiotemporal association rules.
Extracting interesting and useful patterns from
spatial and temporal sets is more difficult than
extracting corresponding patterns from traditional
data due to the complexity of spatial and temporal
data types and spatial relationships changing over
the time.
Our contribution is to process the spatiotemporal
components by computing neighborhood
relationships between geographic objects during a
time interval. This step is achieved by data mining
queries enriched by time management tools. Then
data mining techniques are applied on the resulting
spatiotemporal predicates in order to mine
spatiotemporal association rules.
In the next section we make an overview of the
existing data mining query languages, the section 3
describes our proposed approach of spatiotemporal
association rules mining. Finally, in the section 4 we
summarize the main conclusions of this paper and
point out directions for current and future work.
The high availability of huge databases - rich in
hidden information beyond human’s ability to
retrieve manually- and the prominent necessity of
information and knowledge extraction from such
data, have demanded valuable efforts from the
scientific community. Finding tools and techniques
aiming to analyze these huge data repositories is a
subject dealt by the field of Knowledge Discovery in
Databases (KDD) (Fayyad et al., 1996).
There have been a number of contributions
dealing with different aspects of this problem by
proposing structured languages for KDD
specification. These languages follow SQL patterns
and provide techniques for data preprocessing such
as accessing, cleaning, transforming, deriving and
mining data (Boulicaut and Masson, 2005).
These languages can integrate background
knowledge, like concept hierarchies and can define
thresholds (eg: support, confidence; in the case of
Alouaoui H., Yassine Turki S. and Faiz S..
DOI: 10.5220/0003636303940397
In Proceedings of the International Conference on Knowledge Discovery and Information Retrieval (KDIR-2011), pages 394-397
ISBN: 978-989-8425-79-9
2011 SCITEPRESS (Science and Technology Publications, Lda.)
association rules extraction) in order to extract just
the most interesting patterns (Boulicaut and Masson,
2005). A Data Mining Query Language DMQL has
been proposed in (Han et al., 1996) for mining
association rules using concept hierarchies (Han,
1995) as background knowledge. However, there is
just one practical application of DMQL found in the
DBMiner (Dbminer, 2000) where it is used as a task
description resource.
Another work was proposed by (Meo et al.,
1996) is based on a new operator, named MINE
RULE, designed as an extension of the SQL
language in order to discover association rules.
Other languages have been built on the principles
of relational databases (Imielinski and Virmani,
1999), (Wang and Zaniolo, 2003), (Marcelino et al.,
2004). They follow the SQL patterns with resources
for accessing, cleaning, transforming, deriving and
mining data, beyond knowledge manipulation.
A further important field that needs to get a big
attention is the knowledge discovery from spatial
databases. Complex data types, intrinsic relations
between spatial components and non-spatial
components as well as relationships between data
themselves make the spatial data mining more
difficult. This explains the small number of data
mining query languages that have been proposed for
spatial data. The GMQL (Gegraphic Mining Query
Language) proposed by (Han et al.,
1997) is an
extension of DMQL to support spatial data mining.
Another approach based on the transformation of
a spatial database into an inductive database was
proposed by (Malerba et al., 2004). The proposed
language needs a complex data preprocessing tasks
order to formulate the queries.
A spatiotemporal data mining query language
was proposed in (Chen and Zaniolo, 2000). The
sees reality as instantaneous sequences of
moving objects (Manco et al., 2008), (Bogorny et
al., 2008), (Erlend and Mads, 2010)
and is limited to
mine knowledge from trajectories evolving in space
and time. This language is built on the basis of a
temporal data mining query language proposed by
(Chen and Petrounias, 1998).
All of these languages dealt with traditional,
temporal, or spatial data. They treated separately the
space and the time. The proposed languages merging
space and time aspects were simply limited to the
trajectory of moving objects. To the best of our
knowledge no data mining query language has been
proposed in order to cope with the discrete evolution
of spatial data over the time. Our problematic is to
mine knowledge from discrete evolving objects like
parcels or river changing of shape during large time
intervals. Our proposed queries are settled on a
combination of GMQL and time features.
Our approach aims to extract spatiotemporal
association rules. In order to achieve this objective
two phases should be accomplished; processing the
spatiotemporal components and applying data
mining techniques. Processing the spatiotemporal
components phase is focused on the enrichment of
the spatial association rules extraction with time
concept integration. This means that the mining of
spatial predicates (spatial relationships between
geographic entities such as: close, far, contains,
within....) will be done during time intervals. In
order to accomplish this phase special queries
merging data mining and time management tools
were applied. The extracted spatiotemporal
predicates will be considered as the item-sets treated
by the data mining algorithm Apriori (Agrawal and
Srikant, 1994) in order to mine spatiotemporal
association rules.
Figure 1: The Architecture of the proposed approach.
3.1 Spatial Association Rules
The spatial association rules are extensions of
association rules with spatial features. Associations
highlighted include the proprieties of neighbor
objects and their neighborhood relationship
(Zeitouni and Yeh, 1999). The generated rules have
the form: XÆY where X and Y are sets of spatial
and non spatial predicates. Thus, the spatial
association rules have the following form (1):
1 ∩ … .∩  → 1 .∩  
Extraction of spatiotemporal association rules STAR
STAR application in Risk prediction
Spatiotemporal Data Base
Knowledge data base
Neighborhood relationship computed during time interval
Where at least one of Pi or Qi is spatial. A rule is
always provided with two measures: support (s %)
and confidence (c %). The support is percentage of
transactions that satisfy X and Y among all
transactions from the transaction base.
The Confidence is the percentage of transactions
that verify the conclusion of a rule among those that
satisfy the premise data (Turki and Faïz, 2009).
The determination of the SAR goes through 3
phases (Bogorny, 2006):
Calculation of spatial predicates (spatial
relationships between geographic entities). (Han
et al., 1997); (Koperski, 1999); (Koperski and
Han, 1995).
Generation of frequent itemsets: an itemset is
frequent if at least its support is equal to a
minimum threshold (minsup).
The extraction of spatial association rules.
We noted that the step 2 and 3 of the process of data
mining have received great importance in the
literature and were considered as the major problems
and were designated by frequent pattern mining and
association rule mining. The first step of spatial
relationships calculation is the largest typically
because the effectiveness and efficiency of the
extracted rules is based on these relationships. These
spatial relations are the main characteristic of spatial
data and must be taken into account in the process of
knowledge extraction and this is the primary
characteristic that distinguishes spatial data mining
and classic data mining (Koperski, 1999); (Bogorny,
3.2 The Proposed Query
We have 3 Classes (tables); Road, Town and Oued
(which means a kind of rivers existing in the north
of Africa).
The Town is our target object or (reference
object), the road and the river are its neighbors or
relevant task objects.
The time is stored as an attribute in the table and
we adopted the notion of valid time.
The information in the data base is up-to-date
with the technique of attribute versioning; the new
values or states of an object are stored with the new
interval of validity. The relationships between the
objects are defined on the basis of distance
parameters given by domain experts (e.g: Besides
(Distance > 0m and <= 50m))
Example of the proposed query:
We have two time intervals I1 and I2; I1 ([01-01-
2000, 31-12-2004]) and I2 ([01-01-2005, 31-12-
Q1: Mine spatio-temporal associations
describing Town with respect to
Topology (T.geo, O.geo),
Topology (T.geo, RD.geo),
Tvalidtime_, Ovalidtime, Rdvalidtime
From Town T, Oued O, Road RD
Where distance (T.geo, O.geo) <= “2km”
AND distance (T.geo, RD.geo) <=“500 m”
VALID IN [01-01-2000, 31-12-2004]
The results of these queries are collected and
organized in a knowledge base containing
information related to dependencies between spatial
objects computed during time intervals.
Other neighborhood relationships are computed
and describe some prohibited relations between the
spatial objects.
Example: the relation (oued_contains) is a
vulnerable situation because if an oued (river) covers
an urban zone may possibly cause an inundation.
3.3 Mining Spatiotemporal Association
The computed spatiotemporal predicates were used
in the mining of spatiotemporal association rules
leading to decide about the possibility of natural risk
A set of STAR (Spatiotemporal Association
Rules) was generated.
For example, the following association rule is
derived from the spatiotemporal data set.
oued_touches_I1=yes oued_crosses_I2=yes ==>
oued_contains_I2=yes (40%, 90%)
This rule shows the temporal evolution of the
river (oued) that was near of the town
((oued_touches), (distance > 50m and <=200m)) at
the time interval (I1) then it crossed it
(oued_crosses) at (I2). As a result of this shape
change a prohibited relationship ((oued_contains),
(distance < 0m)) appeared which can be explained
by an inundation risk occurrence.
The values 40% and 90% indicate respectively
the support and the confidence of the rule.
The most meaningful rules will be stored as
learning examples and will be processed by a
learning system (e.g. neural network) in order to
identify unknown future risks. This will be the
object of our future work.
KDIR 2011 - International Conference on Knowledge Discovery and Information Retrieval
In this paper we highlight the necessity to
incorporate the temporal measures in studying the
evolution of geographical objects over the time.
In order to achieve this objective we proposed an
approach aiming to extract knowledge from a
spatiotemporal database by the means of
spatiotemporal mining queries merging both data
mining and time management concepts.
As future work, the most meaningful rules will
be processed by a learning system (e.g. neural
network) in order to identify unknown future risks.
We will also evaluate the possibility to
implement our approach in a wider range of risk
prediction applications including appropriate input
parameters suitable for the study region and the kind
of risk to predict.
Agrawal, R., Srikant, R. 1994. Fast algorithms for mining
association rules in large databases. Research Report
RJ 9839, IBM Almaden Research Center, San Jose,
Bogorny, V., 2006. Enhancing spatial association rule
mining in geographical database. Thesis presented In
partial fulfillment of the requirements for the degree of
doctor in computer science, federal university of” Rio
Grande du sul”, Porto Alegre.
Bogorny, V., Bart, K., Luis, O., 2008. A Spatio-temporal
Data Mining Query Language for Moving Object
Trajectories. Technical Report TR-357. Federal
University of Rio Grande do Sul, Porto Alegre, Brazil.
Boulicaut, J. F. and Masson, C., 2005, Data Mining Query
Languages. In The Data Mining and Knowledge
Discovery Handbook, O. Maimon and L. Rokach
(Eds) (Springer), pp. 715{727.
Chen, X., Zaniolo, C., 2000. SQL
: A Spatiotemporal
Data Model and Query Language. International
Conference on Conceptual Modeling.
Chen, X., Petrounias, I., 1998, Language Support for
Temporal Data Mining. In Proceedings of the
Proceedings of the Second European Symposium on
Principles of Data Mining and Knowledge Discovery
(London, UK: Springer-Verlag), pp. 282{290.
Dbminer Technology 2000.Inc. DBMiner Enterprise 2.0.
Available at DBMiner Technology site. URL: http://
Erlend, T., Mads, N., 2010. Representing topological
relationships for spatiotemporal objects. In
GEOINFORMATICA. (published with open access at
Fayyad, U., Piatesky-Shapiro, G., Smyth, P., 1996. From
Data Mining to Knowledge Discovery: An Overview.
Advances in KDD and Data Mining, AAAI,
Han, J., Koperski, K., Stefanovic, N., 1997. GeoMiner: A
System Prototype for Spatial Data Mining. In ACM-
SIGMOD Int’l Conf. On Management of Data
(SIGMOD'97), Tucson, Arizona.
Han, J., Fu, Y., Wang, W., Koperski, K. and Zaiane, O.,
1996, Dmql: A data mining query language for
relational databases. In Proceedings of the
SIGMOD'96 Workshop on Research Issues in
DataMining and Knowledge Discovery, Montreal,
Canada, pp. 27{33.
Han, J., 1995, Mining Knowledge at Multiple Concept
Levels. In Proceedings of the CIKM (ACM), pp.
Imielinski, T. and Virmani, A., 1999, MSQL: A Query
Language for Database Mining. Data Mining and
Knowledge Discovery, 3, 373{408.
Koperski, K., 1999. A progressive refinement approach to
spatial data mining. Doctorate Thesis. In Simon Fraser
Koperski, K., Han, J., 1995. Discovery of spatial
association rules in Geographic Information
Databases. In proc.4
Int’Symp on large Databases
(SSD’95), pp47-66, Portland.
Manco, G., Baglioni, M., Giannotti, F., Kuijpers, B.,
Raffaet, A., Renso, C., 2008. Querying and Reasoning
for Spatiotemporal Data Mining F. Giannotti and D.
Pedreschi (eds.) Mobility, Data Mining and Privacy.c
Springer-Verlag Berlin Heidelberg .
Marcelino, P., and Robin, J., 2004. SKDQL, a structured
language to specify knowledge discovery processes
and queries. Lecture Notes in Artificial Intelligence
3171, Springer.
Malerba, D., Appice, A. and Ceci, M., 2004, A Data
Mining Query Language for Knowledge Discovery in
a Geographical Information System. In Proceedings of
the Database Support for Data Mining Applications,
pp. 95{116.
Meo, R., Psaila, G. and Ceri, S., 1996, A New SQL-like
Operator for Mining Association Rules. In
Proceedings of the VLDB, T.M. Vijayaraman, A.P.
Buchmann, C. Mohan and N.L. Sarda (Eds) (Morgan
Kaufmann), pp. 122{133
Roshan, N., Asghar, A., 1996. The Management of Spatio-
Temporal Data in a National Geographic Information
System. Springer.
Turki, Y., Faïz, S., 2009. Apport des règles d'association
spatiales pour l'alimentation automatique des bases de
données géographiques. International journal of
Geomatic. Hermès–Lavoisier editions, Paris, France,
Vol. 19/ N°1/2009, pp.27-44.
Wang, H. and Zaniolo, C., 2003, ATLaS: A Native
Extension of SQL for Data Mining. In Proceedings of
the SDM, D. Barbar_a and C. Kamath (Eds) (SIAM).
Zeitouni, K., Yeh, L., 1999. Les bases de données
spatiales et le data mining spatial. International
journal of Geomatic, Vol 9, N° 4.