Data Detection Framework for Spatio-Temporal Data Mining

Data Detection Framework for Spatio-Temporal Data Mining

1P. Madhuri, 2Dr.K. Srinivasa Rao, 3Dr. S.K. Yadav

Department of Computer Science and Engineering1, 2, 3

1Research Scholar, Shri Jagdishprasad Jhabarmal Tibrewala University.

2Professor and Principal, MLRIT, Hyderabad.

3Professor, Shri Jagdishprasad Jhabarmal Tibrewala University.

Abstract : With the unpredictable increase in the generation and use of spatiotemporal data sets, the efficient handling of large volumes of spatiotemporal sets has been the subject of many research efforts. With the phenomenal growth of computer technology everywhere, mining from the enormous amount of spatiotemporal data sets is seen as a central technology that can provide information for real-world applications. Big spatial data apps cover a wide variety of interests, including infectious disease monitoring, simulation of climate change, opioid addiction, and others. As a result, significant research efforts are carried out within these applications to facilitate efficient analysis and intelligence by either offering spatial extensions to existing machine learning solutions or designing new solutions from scratch. Based on the concepts of states and events, the conceptual model was developed and the use of time as a basis for organising spatial data allowed the time and place of any modifications to be recorded. This paper proposes a conceptual-level spatio- temporal modelling approach, called MADS. The idea results from the description of the conditions for a conceptual model to be fulfilled. We can easily formalize spatiotemporal data mining challenges using the proposed knowledge discovery system.

Keywords : Spatio-temporal data mining, discovery of spatiotemporal information, spatio-temporal moving pattern, system for discovery

1. INTRODUCTION

Different technologies for ubiquitous computing and services have been developed in recent times, with wireless sensor networks, context-awareness and open service architecture now considered as the core elements of these programs. The ubiquitous service infrastructure is now being realized in which we can access desired information and services at any time in any area. Most services have a close interaction with real-world artefacts, events, and changes. As such, spatiotemporal information and knowledge are generally required by the services. Many research efforts have been focused on the efficient handling of the large volume of these data sets with the explosive rise in the generation and usage of spatiotemporal data sets. Spatio-temporal sequences. However, there are many difficulties in extracting as well as handling knowledge because of the sophistication and enormity of spatiotemporal information.

The real-time, high performance and continuous extraction of information from large-scale spatiotemporal data sets is needed to provide more intelligent services. Spatiotemporal data mining poses several difficult research problems with regard to the extraction of useful spatiotemporal information from large spatiotemporal data sets.

For example, Ubiquitous services such as telematics, location-based services, surveillance apps, and real-time environment monitoring applications , for example, all require the handling of a large amount of spatiotemporal data sets, but also the promotion of useful information for decision-making in real time. Centered on current techniques for temporal data mining,

From the spatiotemporal data collection, we can discover spatial or temporal knowledge [8] or spatial data mining techniques [4]. However, by considering both the temporal and the spatial elements of objects, new forms of spatiotemporal information for services can be extracted [1, 6, 7].

It is important to endorse a new paradigm for this sort of discovery. First, the discovery question should be more formally specified by any new paradigm for spatiotemporal data mining. In this paper, after considering the several spatiotemporal models previously proposed, we propose a new foundation model for information discovery. Second, based on available and useful spatiotemporal operations and topological relationships, the development of new mining techniques should be encouraged.

Given the remarkable growth of ubiquitous computer technology, mining from the enormous amount of spatio- temporal data sets could become a central technology that can provide information for real-world applications. This paper's description is as follows. We study a number of previous research efforts in section 2 related to the exploration of spatiotemporal information and discuss some of the problems. In section 3, we suggest a 3-tier spatio- temporal information discovery system, and in section 4,

As a base model of the system, we define the spatiotemporal information model. In the 5th segment, The implementation model, which includes a description of the problem and the design of the framework of spatiotemporal discovery of information, is defined.

2. CRITERIA AND RELATED WORK

In the literature, most comparisons of spatio-temporal models usually seek to determine the consistency of a model with regard to its competitors. Therefore, from one article to the next, claims and conclusions differ. There are also a few more general parallels (see, for example, the study of object-oriented temporal languages in [24]). Significant work is recorded in [12,13], which analyses and classifies ten temporal entity-relationship models according to 19 design criteria.

Most of these conditions are peculiar to the temporal dimension. We are interested in the more general question: what characterizes a strong spatio-temporal database conceptual model? Many well-known responses come from previous experience in the design of traditional databases. For example, the word conceptual refers to the ability to provide a clear mapping (i.e. without distortion) between the real world perceived and its representation. In the proposed spatio-temporal models, examples of distortions involve excessive constraints due to weak data structuring capabilities. (e.g. binary relationships only,

no attributes with multivalued values).

Many models (e.g., GeoOM [25] and POLLEN [11]) reflect spatial or temporal characteristics of object types across space or time (e.g., Point, Line, …, Instant, Time- Interval, …). In computational modelling, such artificial object types do not represent

real-world objects of interest, thus contradicting the very first law.

The model has to have strong structures to prevent distortion. Support for entity types, relationship types, multivalued attributes, complex attributes ( i.e. attributes composed of other attributes), is-a links, aggregation (part- of) links, and related integrity constraints is the current norm in this regard (as represented by, e.g., UML[4]). Many more characteristics can be considered, but experience has shown that aiming for the highest expressive power contributes to intolerable difficulty and inevitably results in rejection.

The Model

Also, users are likely to discard models with too sophisticated constructs or with constructs having non- standard semantics. The next important criterion, therefore, is simplicity, which also applies to the visual notations that support schema diagram drawing. Some versions do not contain visual notes [5, 11, 18, 25], whereas others do not contain intuitive notes [10].

Of course, a conceptual model should rely on a sound, formal description and there is a lack of such a sound context in some proposals. Finally, for both data definition and manipulation, an associated data manipulation language would allow the use of the same model.

How the dimensions of space and time are applied to the model is the latest, fundamental problem of spatiotemporal models.

The only correct criterion is, in our view, orthogonality. This idea refers to the independence required between the dimensions of modelling: data structures, space, and time. Database designers should be able to decide the most suitable data structures for

an application, without taking into consideration which spatial or temporal information objects carry information. Once the data structure is fixed, if necessary, space and time features can be freely added. This approach also makes it easier to incorporate space or time characteristics to legacy databases that do not normally contain unique spatio-temporal requirements.

Most of the existing spatial models lack orthogonality, requiring the designer to provide spatial information with a particular representation. GeoOM [25], POLLEN [11], MODUL-R[2], and CONGOO[19], for example, are only

Support the connection of object types with space. An object type may therefore not include attributes that represent spatial

information, such as the reservoir attribute in the Figure 3 example. GéO2[5], on the other hand, does not have any definition of a spatial object; only attributes can be spatial. It is not possible to decide which object type, if any, represents the spatiality of the objects, from the description of an object type with many spatial attributes (e.g., Watershed in Figure 3). Likewise, not every temporal model advocates the correlation of time with artefacts, experiences, and characteristics. In endorsing temporal entity types, for example,[14] lacks the facilities provided with redefinition by conventional inheritance, thereby demonstrating another way of not achieving orthogonality. Beyond orthogonality, model simplicity, as well as ease of use and comprehension, is greatly improved if the modelling of spatial and temporal characteristics relies on similar contexts of reasoning.

The comprehensiveness of a spatio-temporal model is another concern, i.e. how many of the perceived phenomena of space and time should be explicitly capable of representation in the model. This includes problems such as supporting spatiality dimensions of 2.5 or 3 or supporting temporality valid time or transaction time. Without coming to a consensus, several such topics can be debated. The more, as already suggested, is not always the best. Extensible models look like the best answer, but it is not an easy task to determine a model's extensibility. Again, orthogonality proves to be an important quality because the effect of introducing a new idea to the model is restricted by having orthogonal dimensions. However, there is a point we want to emphasize in terms of comprehensiveness. Since we agree that conceptual models should endorse objects, relationships and attributes directly, and we want to include spatio-temporal objects, spatio-temporal relationships, and spatio-temporal attributes since we firmly insist on orthogonality. Present plans lack spatio-temporal relationship support. These relationships have been ignored, depending on the underlying GIS to measure the individual relationships

(mostly topological) from the coordinates of objects stored in the database. An obvious downside of such an approach is that a spatial relationship can not be added to any property,

In conclusion, none of the spatial and temporal models we have studied satisfies any of the above objectives, which contradicts the expressive capacity required in the structural dimension. This prompted the creation of MADS, a conceptual model for object relationship. MADS 's main originality lies in the fact that it is a conceptual model indeed. Still, it has been introduced by translating MADS specifications into usable database models by

We suggest an architecture for the spatiotemporal method of information discovery in order to validate our knowledge discovery model.

Five interrelated layers compose the proposed device architecture: a data layer, an information layer, a spatiotemporal mining layer, a knowledge layer and an application layer.

4. A CONCEPTUAL 3-TIER SPATIOTEMPORAL

KNOWLEDGE MODEL

translators. For example, in [20], a mapping of MADS temporal constructs to TSQL2[23] was defined. For

Information

component

Description

Knowledge

component

conventional land management applications, the MADS model currently covers requirements: one or two- dimensional spatial data, priority for discrete object view versus continuous-field view, valid time only.

Present work on MADS requires an expansion to multi- scale databases and transaction time.

3. ASPATIO-TEMPORAL NOWLEDGE DISCOVERY FRAMEWORK

Data

Spatiotempo ral Informatio n

Spatiotemp

Space Time

Spatiotemporal Knowledge

operation derivation

Fig. 1 Knowledge Discovery Framework A structure is referred to as the collection of principles, models and methodology essential for the exploration of knowledge. As shown in Fig. 1. Our architecture consists of two models: a model of the foundation and a model of implementation.

Here, the foundation model refers to a 3-tier model of spatiotemporal knowledge, which is intended to describe a collection of spatiotemporal objects as well as a volume of knowledge. A spatiotemporal data layer, a spatiotemporal information layer and a

spatiotemporal comprehension layer are included in the conceptual spatiotemporal knowledge model. The representation of data and information, the algebraic operation on spatiotemporal objects and the representation of the knowledge extracted are supported by this model. The spatiotemporal know- ledge discovery model, as an implementation model, involves the concept

of spatio-temporal information and knowledge evolution and the method of discovery of spatiotemporal knowledge.

oral component

object

Space Time Attribute

Fig. 2. Extended Spatiotemporal Knowledge 3-tier model

First of all, to derive useful knowledge the requisite properties of the foundation model over the spatiotemporal phenomenon are as follows. Firstly, the identity and presence of spatiotemporal phenomena must be properly explained. Secondly, the representation of phenomena must be capable of being represented. Thus, not only must the static and dynamic elements of spatiotemporal objects be included, but the temporal element, the spatial element and the attributes must also be considered as the three main elements of the object. Third, for the study of phenomena, it must be capable of reflecting multilevel context information. Finally, it must be able to use previous experience to mine new forms of knowledge.

Among the current knowledge models, we chose the Pyramid model of Mannis [5] as the most fitting model of spatiotemporal knowledge and developed it to meet our own requirements.

The computational knowledge model proposed for the replication of spatiotemporal phenomena actually reflects the spatiotemporal data observed, the spatiotemporal information analyzed, and the spatiotemporal knowledge discovered. As shown

in Fig. 2. In this model, the data portion consists of representations of the spatiotemporal data occurring in the spatiotemporal universe. In order to extract the value, the information component concerns the selected and

meaningful data extracted from the data component by using a generalization or transformation operation. Finally, the aspect of knowledge describes objects of knowledge which are useful to human beings. Awareness may be explicitly induced or implicitly induced from spatiotemporal

information or a priori awareness from a collection of spatiotemporal items. Our conceptual knowledge model for representing spatiotemporal phenomena, focused on the idea of a value chain and conceptualization, clearly represents the spatiotemporal data observed, the spatiotemporal information processed, and the spatiotemporal knowledge discovered. Compared to Manni's two-tier data component and knowledge model, our model is able to abstract the various operations performed on the spatiotemporal data set by providing an information component. The abstraction of spatiotemporal activities makes it possible for the current legacy DBMS systems such as relational DBMS and OR DBMS to incorporate the information discovery framework.

5. SPATIOTEMPORAL KNOWLEDGE DISCOVERYMODEL

Definition of Spatiotemporal Knowledge

To better represent the spatiotemporal information mentioned above, it is important to include a method of representing temporal qualifications in the temporal domain. A more complete description of the time expression of temporal information has been given by Lee [2], including the time expression defined by Chen. The calendar interval expression set is denoted as CIS for each time expression set, the calendar relationship expression set as CRS, the time interval expression set as TIS, and the periodical time expression set as CIS. PTS as well. Let TQ Timexpo be the temporal qualification, where Timexpo={CIS, CIS, TIS, PTS}. In addition, the manner in which the spatial qualification of the spatial domain is interpreted Has to be given The spatial qualification SQ of knowledge applies to the spatial domain ES, where it refers to a valid spatial domain coverage of knowledge A primitive RP of spatiotemporal information is an indivisible information unit with a temporal qualification, as well as a spatial qualification along with a definition of the law. Since primitive spatiotemporal knowledge can be expanded by adding temporal and spatial qualifications from the existing definitions of the law, knowledge can be described as follows. Let primitive RP be spatiotemporal information < Rule, TQ, SQ >. Now, this concept will redefine existing forms of information,

such as descriptive information, temporal knowledge and spatial knowledge. Spatiotemporal knowledge R is defined as a compound of the primitives of spatiotemporal knowledge and as follows, the unique identifier Rid. R = < Rid, ri + >, where ri +, where ri RP. As a set of information discovered, i.e. {Ri Ri R}, a spatiotemporal information set RS is specified.

In the timeline T, information was located at a time point t.

A new spatiotemporal information snapshot set is created whenever a discovery occurs. Spatiotemporal knowledge discovery The discovery of spatiotemporal knowledge consists of the discovery of

a spatiotemporal knowledge set from the spatiotemporal data set given. It should also be possible for spatiotemporal awareness to link space and time with object and relationship types in order to convey an intrinsic spatiality or temporality that does not depend on their semantic attributes. A predefined feature, geometry, MADS, allows object and relationship instances to be suspended and resumed in their membership to describe the spatiality of an object or relationship form. The status attribute therefore differs in time: it is a DBTIME to STATUS function, where STATUS is a predefined domain consisting of four values: non-existent, active, suspended, and disabled.

The first value relates to an object that is believed to exist sometime later. An object that can be used within the associated timeframe is defined by the active value. It is possible to read but not change the properties of suspended objects. Finally, the disabled attribute identifies an object that has existed in the past but is not available during the disabled timeframe [9]. The following restrictions apply to transformations between values: not-yet-existing-> active, active <-> suspended, active-> disabled, and suspended-> disabled. Status-related features include the life cycle (the birth-to-death interval) and the active cycle (the active status-related set of intervals / instants).

6. SPATIO-TEMPORAL KNOWLEDGE

DISCOVERY

The discovery of spatiotemporal knowledge consists of the discovery of a spatiotemporal knowledge set from the spatiotemporal data set given. The problem of extracting the spatiotemporal information set RS from all information R= < Rid, ri+ > that meets the given thresholds from the given spatiotemporal data set DS, using the context knowledge provided, is known as a spatiotemporal knowledge discovery. Two flows are composed of the proposed information discovery process model. In other words, the development of a mining method and the discovery of spatiotemporal information. The creation of the discovery objective, data discovery, a summary of the discovery profile, and the development of a mining operation are the key development processes of information discovery techniques. Information discovery processes consist of five steps: task creation, specific data set creation, spatiotemporal data mining, visualization, and analysis and assessment.

.

System Architecture

MPMine()

)

MPMine()

)

Spatiotemporal Application Support Layer

(Mining Control & APIs)

Spatiotemporal

Application

Spatiotemporal

Application

MPMin

Spatiotemporal Mining Layer

Spatiotemporal Knowledge Layer

Moving Pattern

Domain

Knowledge

Moving Pattern

Domain

Knowledge

(Rule Management

Spatiotemporal

Knowledge

Spatiotemporal

Knowledge

(Rule Mining Modules) Modules)

MPMine()

MPMine()

Knowled ge

Knowled ge

Spatiotemporal Data & Information Layer

SCGeneraliz

Spatiotempal or Mining

Spatiotemporal

Information

Spatiotemporal

Information

(TDBMS, MODB, GIS, DBMS) MP SCH

Fig. 3. Knowledge discovery system architecture based on the framework Just in Fig. 3,

Spatiotemp oral

Spatiotemp oral

Informati on

ContainInRegi ( TimeExp(

The key components and their relationships are represented within our implementation model. Entry to the heterogeneous spatiotemporal data set helps the spatiotemporal data and knowledge layer. In this layer, the permission request is mapped to the physical access of each storage from the higher layer. The spatiotemporal mining layer performs the extraction and transformation of data, and then uses spatiotemporal mining operators to discover rules. Therefore, in this layer, access to the lower layer must be supported. Both the application layer and the mining layer are connected with the information layer. Novel features such as the representation of rules, the access and preservation of valid rules and the translation of the rules to applications are included in this layer. Implementation and discussion Implementation of a Spatiotemporal know- ledge discovery system.

To demonstrate how the proposed structure can be extended to the creation of a system for the discovery of spatiotemporal information, we briefly define the system for the discovery of spatiotemporal moving patterns, i.e. MPMiner, which is implemented based on the proposed architecture as a semi-tightly coupled structure. As mentioned above, the five layers in the framework's implementation model correspond to the framework 's three layered conceptual spatiotemporal information model.

These major modules consist of the framework. As seen from Fig. 4, MPMiner is a method of moving pattern discovery that includes the MPMine) (operation as the main mining operation and is introduced as a semi-tightly coupled structure. Via an extension of the DBMS procedure and the implementation of the information and knowledge portion of the proposed architecture, this structure has the advantages of successful data processing.

M MA

Data Compon

Data Compon

Fig. 4. Implementation of MPMiner the implementation of this scheme is split into the implementation of the system and the execution of the UI and mining activities. The implementation of the system correspnds to the implementation of the conceptual model of the three elements. The data portion is implemented using Oracle 8i as a set of relationships. Just in Fig. 4, contains the MD data collection of moving objects and the Chart of context map data. The main relational algebra activities are expressed as SQL, as the knowledge portion. Using PL / SQL, spatial and temporal operations are represented as the stored procedure. It contains the Timexpo) (function for time expression processing and the SC Generalize) (function. The knowledge portion is implemented as a set of relations in Oracle 8i in order to represent the discovered knowledge and context knowledge. HG-Trees are often implemented as a collection of connections by normalization to represent a spatial conceptual hierarchy. Also, the spatiotemporal moving patterns discovered are stored as a relationship.

DISCUSSION

As a result of the assessment, the proposed paradigm provides the proper representation of the spatiotemporal phenomena for knowledge discovery by providing a foundation model that is based on the event-based model and that hierarchically represents the three related components such as data, information , and knowledge. Some concerns still occur, however, where data limitations and confidentiality are not taken into account. The system supports forms of multiple knowledge with regard to context knowledge, such as temporal knowledge, spatial knowledge and spatiotemporal knowledge, as well as user- defined knowledge. The method of representing knowledge, however, and the kinds of knowledge operations are still to be included. It promotes integrated information exploration with regard to the support of knowledge discovery integrated with the already knowledge minded. Maintaining knowledge and the evolution of knowledge, however, has not been properly considered Finally, even operation abstraction is possible with regard to implementation in the legacy system, although in some legacy systems it is limited. The system

needs to be strengthened in the future by considering the concerns that have been found.

7. CONCLUSION

The emergence of ubiquitous computing, the development of wireless sensor network technology, and the growing need for context-aware services and intellectualization have led to increased understanding of the enormous volume of spatio-temporal data and the study of information discovery. Many of the new services strive to provide consumers with more customized and intelligent features. Real-time, high performance and continuous extraction of information from the enormous amount of spatiotemporal data sets is required to provide more bright services. In this paper, we discuss a structure for the exploration of spatiotemporal information that encourages the creation of new types of knowledge, such as the pattern of spatiotemporal movement. The meaning and relationships of spatiotemporal data sets and information can be interpreted in the proposed system by using a foundation model for knowledge discovery. In addition, the concepts of spatiotemporal knowledge and knowledge discovery processes and models are There are some merits to the spatiotemporal information exploration paradigm discussed here. However, given the characteristics of the spatiotemporal data sets of recent applications, the size of the data sets is growing because of the continuously evolving spatiotemporal artefacts, because of the convergence of services, the need for real-time analysis has increased, and the semantics of derived information are becoming more complicated.

8. REFERENCES

[1] T. Abraham, Knowledge Discovery in SpatioTem- poral Databases, School of Computer and Infor- mation Science, University of South Australia, Ph. D dissertation,1999.

[2] Lee Y.J., Data Mining Technique for Discovering Temporal Relation Rules, Department of computer science, chungbuk national university of korea, Ph. D dissertation, 2001.

[3] Jeong J.D., Paek O.H., Lee J.W., and and Ryu K.H., Temporal Pattern Mining of Moving Object for Location-Based Service, In Proc. of International Conference on Database and Expert Systems Applications (Dexa2002), (LNCS2453), 2002.

[4] K. Koperski, J. Han, and J. Adhikary, Mining knowledge in geographical data, to appear in Communications of the ACM, 1998.

[5] J. Mennis, and D.J. Peuquet, A Conceptual Frame- work for Incorporating Cognitive Principles into Geographical Database presentation, International Journal of Geographical Information Science,Vol. 14, No. 6, pp. 501-520, 2000.

[6] Lee J.W., Spatiotemporal Moving Pattern Discovery Technique based on Knowledge Discovery Frame- work, Department of computer science, chungbuk national university of korea, Ph. D dissertation, 2003.

[7] J.F. Roddick and M. Spiliopoulou, Temporal data mining: survey and issues, Research Report ACRC- 99-007, University of South Australia, 1999.

[8] S.A. Sarabjot, D.A. Bell, and J.G. Hughes, The role of domain knowledge in data mining, In Proc. of the Int. Conf. on Information and Knowledge Manage- ment, pp. 37-43,1995.

[9] Tsoukatos and D. Gunopulos, Efficient Mining of Spatiotemporal Patterns, In Proc. of the 7th Int. Symp. on Spatial and Temporal Databases (SSTD), 2001.

[10] E. Mesrobian, R.R. Muntz, E.C. Shek, J.R. Santos, J. Yi, K. Ng,

S.Y. Chien, C.R. Mechoso, J.D. Farrara, P. Stolorz, and H. Nakamura, Exploratory Data Mining and Analysis Using Conquest, IEEE Pacific Rim Conference on Communications, Computers, Visualization, and Signal Processing, May, 1995.

[11] R.T. Ng and J. Han, Efficient and Effective Clustering Method for Spatial Data Mining, In Proc. of International Conference of Very Large Data Bases, pp. 144-155, 1994.

[12] R.T. Ng, Spatial Data Mining: Discovering Know- ledge of Clusters from Maps, In Proc. of ACM SIGMOD Workshop on Research Issues on Data Mining and Knowledge Discovery, 1996.

[13] R.E. Valdes-Perez, Systematic Detection of Subtle Spatio- Temporal Patterns in Time-Lapse Imaging. I. Mitosis, Bioimaging. Vol. 4 , No. 4, pp. 232-242, 1998.

[14] J.F. Roddick and B.G. Lees, Paradigms for Spatial and Spatio- Temporal Data Mining, Geographic Data Mining and Knowledge Discovery. Taylor and Francis. Research Monographs in Geographic Information Systems. Miller, H. and Han, J., Eds, 2001.

[15] H. Gregersen, C.S. Jensen, and L. Mark. Evaluating Temporally Extended ER Models. In Proceedings of the 2nd CaiSE97 International Workshop on Evaluation of Modeling Methods in Systems Analysis and Design, 1997.

[16] R.M. Guting et al. A Foundation for Representing and Querying Moving Objects. FernUniversitaet Hagen,

[17] Informatik-Report 238, September 1998

[18] IEF Information Engineering Facility, A Guide to Information Engineering Using the IEF, Texas Instruments, 1988.

[19] S.F. Keller, A Presentation Model and a Mapping Language for INTERLIS, Document RFC-1011e, Federal Directorate of Cadastral Surveying, Berne, Switzerland, September 1997.

[20] M.R. Klopprogge and P.C. Lockeman. Modeling Information Preserving Databases: Consequences of the Concept of Time. In Proceedings of the 9th International Conference on Very Large Databases, VLDB'93, 1983.

[21] D. Pantazis, J.P. Donnay. La conception de SIG РM̩thode et formalisme, Herm̬s, 1996.

[22] C. Parent, S. Spaccapietra, E. Zimányi, Modeling time from a conceptual perspective. In Proceedings of the 7th International Conference on Information and Knowledge Management, CIKM'98, p. 432-440. 1998.