A Management Model of Real-time Integrated Semantic Annotations

to the Sensor Stream Data for the IoT

Besmir Sejdiu

, Florije Ismaili

and Lule Ahmedi

Contemporary Sciences and Technologies, South East European University, Tetovo, Macedonia

Faculty of Electrical and Computer Engineering, University of Prishtina, Prishtinë, Kosovo

Keywords: Internet of Things, Wireless Sensor Networks, Stream Data Management, Semantic Annotations, Water

Quality Monitoring.

Abstract: Wireless Sensor Networks (WSNs) are one of the most important components of the Internet of Things (IoT).

They produce continuous stream of data and transmit these data to a centralized server. Due to the dramatic

increase of streaming data, their management and exploitation has become increasingly important.

Furthermore, by adding semantic annotations into sensor stream data, better understanding and more

meaningful descriptions is provided, which enables application areas of IoT to become much more intelligent.

In this paper, a data stream management model of WSNs for IoT real-time monitoring systems, that supports

real-time integration of data from heterogeneous sensors with semantic annotations is presented. To validate

the proposed model, an IoT system for real-time water quality monitoring is built, which enables real-time

integration of semantic annotations to the sensor stream data in the format of Sensor Observation Service

(SOS).

1 INTRODUCTION

Nowadays, smart water networks, smart homes, smart

cities, smart health, smart grid, intelligent

transportation, are infrastructure systems that connect

our world more than we ever thought possible. The

common vision of such systems is usually associated

with one single concept, the Internet of Things (IoT)

(Yinbiao, 2014). The idea of IoT was developed in

parallel to Wireless Sensor Networks (WSNs).

Therefore, the main components that enable IoT are

WSNs (Lazarescu, 217).

A Wireless Sensor Networks is a wireless network

consisting of spatially distributed autonomous

devices using sensors to monitor physical or

environmental conditions. WSNs may be

homogeneous or heterogeneous. Homogeneous

sensors send only one type of information (e.g. the

water temperature), while heterogeneous sensors

send more than one type of information (e.g.

temperature and dissolved oxygen). All these sensors

send observational data referred to as sensor stream

data to a remote server. Therefore, sensory data

https://orcid.org/0000-0002-2786-5384

https://orcid.org/0000-0002-3627-0147

https://orcid.org/0000-0003-0384-6952

comes from multiple sensors of different modalities

in distributed locations.

Furthermore, sensor stream data is enabled to the

web through the Sensor Web (SW). SW by

incorporating technologies of the Semantic Web

creates the Semantic Sensor Web (SSW) (Wang,

2020). Therefore, by adding semantic annotations to

sensor stream data with concept definitions from

domain knowledge (e.g. ontologies), the

interpretation and understanding of sensor data and

metadata is enabled. The real-time integration of

sensor data as dynamic data with semantics is defined

as real-time semantic annotation, while sensor data

that is stored in repository (data store) as static data,

and then integrated with semantics is defined as non-

real-time semantic annotation (Sejdiu, 2020).

Organizations like Open Geospatial Consortium

(OGC) and World Wide Web Consortium (W3C)

have proposed industry standards such as Sensor Web

Enablement (SWE), which are aimed at providing

unified standards (Shi, 2018).

To manage the real-time integration of semantic

annotations into heterogeneous sensor stream data

Sejdiu, B., Ismaili, F. and Ahmedi, L.

A Management Model of Real-time Integrated Semantic Annotations to the Sensor Stream Data for the IoT.

DOI: 10.5220/0010111500590066

In Proceedings of the 16th International Conference on Web Information Systems and Technologies (WEBIST 2020), pages 59-66

ISBN: 978-989-758-478-7

with context in the IoT and by using OGC SOS

standards, we introduce a management model in this

paper.

The paper is organized as follows: Section 2

provides a discussion on related work for semantic

annotations to the sensor stream data. Section 3 is an

overview of the data stream models, annotated sensor

data streams, and used technologies. Section 4

represents a proposed management model of

integrated semantic annotations to the sensor stream

data for the IoT. Prototype implementation of the

proposed model is presented in Section 5, and system

outputs is presented in Section 6. Finally, Section 7

concludes the paper and relevels some of the future

perspectives of the proposed model.

2 RELATED WORK

Recently, some researchers have already shown up

with several investigations related to semantic

enrichment of sensor stream data. Lin et al. (Lin,

2019) proposed a semantic annotation method to

annotate IoT sensor data through semantics. They

used K-means clustering method for knowledge

discovery. The proposed mechanism is

semiautomatic, because after clustering the data,

there are just clusters without knowledge. The

knowledge of the clusters is defined by people.

Xiaomin el al. (Xiaomin, 2016) proposed an approach

for ontology modelling which is used in evaluating

the river water quality and its relevant processing

knowledge. The presented model consist of the data

acquisition layer, diagnosis layer and decision

support layer.

Furthermore, in this study, different solutions are

identified, (Vera, 2014), (Pradilla, 2016), (Bytyçi,

2019), to semantically annotate sensor stream data.

Those proposed solutions used non-real-time

semantic annotation because the sensor stream data is

stored in data store or ontology as static data and then

integrated with semantics. However, it is still an open

issue on how techniques and models for integration

and interpretation of the semantic annotations in real-

time in the sensor stream data should be advanced.

The main contributions of our approach are

presented as follows:

 A data stream management model of WSNs for

real-time IoT monitoring systems has been

developed, which support real-time integration

of data from heterogeneous sensor with

semantic annotations.

 To validate the proposed model, an IoT system

for real-time water quality monitoring is built,

which enables real-time integration of semantic

annotations to the observational data on water

quality coming from wireless sensors.

 By incorporating OGC SOS standards, the

heterogeneity of data sensor is hidden from

arbitrary sources, and a real-time service

interface for published enriched sensor stream

data is provided with the semantic annotations

to display in IoT real-time monitoring systems.

3 BACKGROUND

3.1 The Data Stream Models

Depending on characteristics, source of data

transmission and saving of stream data, those can be

modelled in various ways: Real-time data stream

(data stream in real time is a sequence of data which

arrives in order and/or in pre-processed ways),

Stream items (since data is received in streams, those

can be modelled as a sequence in a list of elements),

and Window models (only a fragment of the streaming

data is of interest at any time, and can be classified

according to the three criteria: Fixed sliding window,

Landmark window, and Adaptive window).

3.2 Annotated Sensor Data Streams

WSNs consist of small-scale devices that enable

observe various physical phenomenon, which provide

sensor data in raw format. Typically, classical IoT

applications cannot interpret the sensor data and

understand its context. This makes it nearly

impossible to get the high-level information of the

events and infer additional knowledge to gain

situational awareness (Khan, 2015).

To provide meaning (semantics) of raw data,

annotated sensor data stream is required. The

annotated sensor data becomes more meaningful and

understandable, enabling end-users to get high-level

details about the real-world situations instead of raw

sensor data. This is known as Semantic Sensor Web

(SSW). These annotations provide more meaningful

descriptions and enhanced access to sensor data than

SWE alone, and they act as a linking mechanism to

bridge the gap between the primarily syntactic XML-

based metadata standards of the Sensor Web

Enablement (SWE) and the RDF/OWL-based

metadata standards of the Semantic Web.

To encode semantic annotations and data

gathered by sensors, SWE is used in this paper,

respectively version 2.0 of the Sensor Observation

Service (SOS) standard relies on the OGC

Observation & Measurement (O&M).

WEBIST 2020 - 16th International Conference on Web Information Systems and Technologies

3.3 Technologies

The proposed management model of real-time

integrated semantic annotations to the sensor stream

data for the IoT utilizes:

Spark Streaming

- is an extension of the Apache

Spark which enables to build scalable fault-tolerant

IoT applications. Data can be ingested from many

sources like Apache Kafka, TCP sockets, Flume,

Twitter, etc., and processed data using complex

algorithms expressed with high-level functions like

map, reduce, join and windows. Finally, processed

data can be pushed out to file systems, databases, and

live dashboards.

Apache Kafka

- is a distributed streaming

platform which has capabilities to publish and

subscribe to streams of records, similar to a message

queue or enterprise messaging system.

Apache Cassandra database

- is a distributed

store for structured data that scale-out on cheap. It is

designed to handle large amounts of data.

4 PROPOSED MODEL

In Figure 1, an overview of the model architecture for

processing and managing sensor stream data for IoT

real-time monitoring systems, such as: water quality

monitoring, air quality monitoring, etc. is presented.

The WSNs are deployed in different locations.

They produce a continuous stream of data, and

transmit to Apache Kafka in various formats (e.g.

binary, JSON, XML, etc). Kafka is utilized to

transform them in a specific format that will be

processed by Spark Streaming in real-time and

parallel. The Spark Streaming enables a real-time

integration of semantics into sensor stream data by

using association rules, mining data streams, WSNs

metadata, and archival data streams, with concept

definitions from ontologies or other semantic sources,

which provides the understanding and more

meaningful descriptions to enable application areas of

IoT to become much more intelligent.

The enriched sensor stream data with semantic

annotations results are stored in the Cassandra

database, and will be displayed in IoT real-time

monitoring systems in format of SOS O&M standard

by using stakes, such as XLink (without including

XPath). A fragment of an output example is

presented in Figure 2.

As shown in Figure 1, the proposed data stream

management model support real-time data integration

of heterogeneous sensors with semantic annotations,

http://spark.apache.org

https://kafka.apache.org

continuous queries on streaming data, outlier

validation of streaming data, ad-hoc queries, and

archive stream data with their semantic annotations

for applications that need to answer queries from the

archival store (persistent data stored). The proposed

model consists of the main components: A) Input

Data Stream, B) Stream Processor, C) Data

Modelling, D) OGC standards, and E) Ontology.

Each component of the model is described in

details as following:

A. Input Data Stream – is implemented in Apache

Kafka and accepts in real-time input streaming data

sent by the WSNs. Unordered streams come without

any kind of pre-processing – as unordered cash

schedule, they do not need to have the same data rates

or data types, and the time between elements of one

stream need not be uniform.

B. Stream Processor is developed in Spark

Streaming and contains Outlier Stream Validator &

Classificator, Query Process, Ad-hoc Queries, and

Semantic Annotations Stream Process:

• Outlier Stream Validator & Classificator – is

part of the stream processor in charge of real-time

validating sensor streaming data, which marks stream

data with one of the following status, ‘valid’ or

‘outlier’. Validated data continues to processing

forward, while invalid data are stored in Invalid Data

Streams (IDS). A data stream object is considered an

outlier if it does not conform to expected behaviour,

which corresponds to either noise or anomaly. For

example the observed value of the pH sensor is ‘-2’

or ‘NULL’, in this case this sensor data will be

classified as outlier because the range value of pH

phenomena is 0 to 14. Outliers can arise due to

different reasons such as mechanical faults, other

changes in the system, fraudulent behaviour,

instrument error, human error or natural deviation

(Yu, 2020). Therefore, the Outlier Stream Validator

& Classificator provides data quality for IoT real-

time monitoring systems.

• Query Processor – supports continuous queries

for streaming data, and are continuously executed as

data streams continue to arrive. The answer of the

continuous query is produced over time, always

reflecting the stream data seen so far. The answers of

our Query Processor can include semantic annotated

data in the result.

• Ad-hoc Queries – queries executed ad-hoc from

users; a question asked once about the current state of

a stream or streams. Users can also specify ad-hoc

queries that integrate streaming data and persistent

data stored on Working Data Streams,

http://cassandra.apache.org

A Management Model of Real-time Integrated Semantic Annotations to the Sensor Stream Data for the IoT

IoT Real-Time

Monitoring Systems

Sensor Stream Data

SId:1;Param:Temperature;

Value:17.15;Lat:42.706703186;

Long:21.038431;

Timestamp:20200312165213

Archival

Data

Streams

(ADS)

. . .

Query

Processor

Working Data

Streams

(WDS)

Invalid Data

Streams

(IDS)

SId:2;Param:pH;

Value:5.68;Lat:42.706703186;

Long:21.038431;

Timestamp:20200312165423

SId:3;Param:DissolvedO.;

Value:47.20;Lat:42.706703186;

Long:21.038431;

Timestamp:20200312165933

Semantic Annotations

Stream Processor

WSNs

Metadata

(WMD)

Working Data

Streams

Annotations

(WDSA)

Archival

Data

Streams

Annotations

(ADSA)

Stream Processor

OGC

Standards

WSNs

Apache Cassandra Database - Data Modelling

Processor

Data Streams

(PDS)

Ad-hoc

Queries

Ontology

Figure 1: An overview of the model architecture.

WSNs Metadata or Archival Data Streams. In

addition, the result of Ad-hoc Queries can include

semantic annotated data in the result.

• Semantic Annotation Stream Processor – is part

of the Spark Streaming processor which enable a real-

time integration of semantic annotations into

heterogeneous sensor stream data with context in the

Internet of Things. This component can use sensor

metadata, archival data streams, mining data streams,

association rules (with concept from ontologies), or

other sematic sources for adding semantic

annotations to the sensor stream data. The semantic

annotated of sensor stream data are stored in Working

Data Stream Annotations.

C. Data Modelling is developed in Apache

Cassandra database and contains: Processor Data

Streams, Working Data Streams, Working Data

Stream Annotations, Archival Data Streams, Archival

Data Stream Annotations, Invalid Data Streams, and

WSNs Metadata:

• Processor Data Streams (PDS) – contains a

summary of streaming data for Stream Processor,

which can be used for answering queries. For each

deployed sensor, only a row is saved which includes:

Sensor Id- an identifier that uniquely identifies a

sensor. Sensor Parameter - name of the parameter or

phenomena that sensor measure (e.g. temperature,

humidity, pH, etc). Sensor Current Value- current

measured value from the sensor. Sensor Total Rows -

the total number of measurements by the sensor since

its deployment. Sensor Max Value - the maximum

value measured by the sensor since its deployment.

Sensor Min Value - the minimum value measured by

the sensor since its deployment. Sensor Sum Value -

the sum of values measured by the sensor since its

deployment. Sensor Avg Value - the average value

measured by the sensor since its deployment, which

is derived by finding quotient between sum of values

measured by the sensor and the total number of

measured by sensor (Sensor Sum Value/ Sensor Total

Rows). Sensor Window Max - the maximum value of

sliding window which contains last n values, where n

is a configurable number (e.g. 15 last measured

values) sent by the sensor. Sensor Window Min - the

WEBIST 2020 - 16th International Conference on Web Information Systems and Technologies

minimum value of sliding window. Sensor Window

Avg - the average value of sliding window. Sensor

Current Timestamp - current measured timestamp

sent by the sensor. Sensor Current Latitude &

Longitude - current latitude and longitude (current

geographical position from where sensor has sent

data).

• Working Data Streams (WDS) – contains

streaming data for operation of Stream Processor,

which are configurable according the quantity and

can be used for answering queries. So, it’s a Fixed

Sliding Window that contains the last sensor

streaming data (e.g. 15 last measured values - its

configurable number). For each measured values

information is stored as described: Observation Id -

an identifier that uniquely identifies the observation

in the WDS. Sensor Id - an identifier that uniquely

identifies a sensor. Sensor Parameter - name of the

parameter or phenomena (e.g. temperature) measured

by the sensor. Sensed Value - measured value that is

sent by the WSN. Timestamp - time when the sensed

value has been generated by WSN. Latitude &

Longitude - geo location, geographical position from

where the sensor has sent data. It is especially useful

when a sensor is attached to a moving object such as

a car, airplane, etc. or in case of mobile sensor to

perform monitoring in different ad-hoc selected

locations of interest, while in case of static sensor that

perform monitoring in the region of interest, the geo

location can be NULL because the location of these

senor type can be saved as metadata in WSNs

Metadata. Observation Id - a code which identifies a

single sensor node measurement, example in cases

where a sensor node measures three parameters, e.g.

temperature, turbidity and conductivity at the same

time (as a single measurement), then all of three

measurements will get the same transaction code,

otherwise the transaction code will be NULL. Entry

Timestamp - date and time when streaming data has

arrived to Stream Processor.

• Working Data Stream Annotations (WDSA) –

stores semantic annotations of sensor streaming data.

The Semantic Annotation Stream Processor

component, is used to tag in real-time integration the

sensor data stream with semantic annotations. One

measurement that is stored in Working Data Streams

can obtain several sematic annotations, which

includes information: Annotation Id - an identifier

that uniquely identifies a sematic annotation.

Observation Id - references to Working Data Streams

observation Id. Annotated Date - date and time when

sensor streaming data has been tagged with sematic

annotations. Annotated Type - represents the type of

annotation, ‘Embedded’ (only a single value-scalar of

semantic annotation) or ‘External’ (an external

resource linked by ‘XLink’ that point to our ontology

‘ont-core.owl’). Annotated Value - stores the

semantic annotated value. Example value of

‘Embedded’ annotation type for water status can be

‘Good’, ‘Moderate‘, ‘Poor’, ‘Bad’ or ‘High’:

Example value of ‘External’ annotation type can be

(for more details see Figure 2):

<annotation

xlink:href="http://myserver/ontologies/

ont-core.owl#WaterStatus_ClassV"/>

• Archival Data Streams (ADS) – archives data

streams for generating reports and different statistics.

The structure of data modeling of ADS is the same as

WDS.

• Archival Data Stream Annotations (ADSA) –

archives semantic annotations of sensor stream data

for generating reports and different statistics. The

structure of data modeling of ADSA is same as WDSA.

• WSNs Metadata (WMD) – the data describing

wireless sensor networks itself, its devices and the

corresponding site allocation data. This data is named

as static data that describes the wireless sensor

networks in the field, its configuration which might

involve node types, like sensing nodes, gateway

nodes, central monitoring node, and description of

sensors as devices (sensor name, serial number,

manufacturer, and type), as well as data about the

deployment sites, like sensors’ location, example for

water system monitoring, the river basins,

municipalities the rivers belong to, etc.

• Invalid Data Streams (IDS) – archives invalid

sensor stream data that is classified as outlier by

Outlier Stream Validator & Classificator. The data

stored in IDS is optional and it’s depends on the

system requirements.

D. OGC Standards - as mentioned above, the

enriched sensor stream data with the semantic

annotations results will be published to IoT real-time

applications in format of OGC standards, respectively

version 2.0 of the SOS O&M standard.

E. Ontology - an ontology named ‘ont-core.owl’

is created, see Figure 3. Here are developed semantic

annotations for international regulatory of water

quality, such as:

 United Nations Economic Commission for

Europe (UNECE) with semantic annotations of

water status: Class I, Class II, Class III, Class

IV, and Class V.

 Water Framework Directive (WFD) with

semantic annotations of water status: Good,

Moderate, Poor, Bad, and High.

Details about the working cycle of this model are

as follows: streaming data is sent by the wireless

sensors networks in Input Data Streams (Apache

A Management Model of Real-time Integrated Semantic Annotations to the Sensor Stream Data for the IoT

Kafka). Sensor stream data is an array of different

types containing sensor id (sid), name of the

parameter, measured value of the sensor,

geographical position (latitude and longitude) and

timestamp, as seen below:

‘SId: 1; Parameter: Temperature;

Value: 17.15; Lat: 42.706703186; Long:

21.038431; Timestamp: 20200312165213’

Then, the validation of the stream data elements

occurs through Outlier Stream Validator, in which

every sensor stream data takes the status of validity

(true – data is valid or false – data is outlier). When

data takes the validation status ‘true’, it will be

transmitted for further processing in Semantic

Annotated Stream Processor which makes real-time

integration of semantic annotations into those stream

data. Then the enriched sensor stream data with the

semantic annotations results will be stored in WDS

and WDSA and it will be transformed in SOS O&M

format to displayed in IoT real-time monitoring

systems.

It is worth mentioning that when a new value is

measured by the sensor, it arrives in WDS (and their

semantic annotations are stored in WDSA) then the

oldest value is removed from there and goes to ADS

(respectively ADSA) for archiving. So, in ADS and

ADSA are archived data which serve for generating

reports and statistics for longer time frames.

5 PROTOTYPE

IMPLEMENTATION OF

PROPOSED MODEL

To validate the proposed model in this study, we

implemented a prototype illustrating a water quality

monitoring use case, named as Water Quality

Monitoring System (WQMS). This system enables

the measurement of water quality in real time by

applying the latest technology trends, such as wireless

sensor networks, which provide continuous

monitoring, and consist of nodes referred to as motes

that are sensitive to the environment where they are

deployed.

WQMS enables water parameters monitoring

such as: temperature, potential of hydrogen (pH), and

dissolved oxygen (DO). The type, rank, and unit of

these parameters are shown in Table 1. The data

produced by the sensor is a real number, e.g. 16.5°C

for temperature measurement. The sensors are

configured in such a way that each node will send data

every 10 minutes. So the rate of the data stream is not

high, therefore the storing of these data by applying

the proposed model, nowadays, will not be a problem

for the new technologies.

Table 1: Specification of water parameters.

Parameter

name

Parameter Type

Range

value

Unit

Temperature

Thermal

conditions

-1 to

+50

°C

pH Acidification 0 to 14 pH

Dissolved

Oxygen

Oxygenation

conditions

0 to 300 %

The WQMS system architecture is presented in

Figure 4. It mainly consists of the static wireless

sensing nodes, mobile wireless sensing nodes,

gateway node and the monitoring node. Static

wireless sensing nodes are located in a given position

and through gateway node, continuously transmit

sensed data to the central monitoring node, while

mobile wireless sensing nodes can move from one

position to the other to measure water parameters.

The sensor data in the central monitoring node are

transmitted via 3G/GPRS using web services. The

WQMS software is installed in the central monitoring

node, in which is implemented the proposed

management model.

6 SYSTEM OUTPUTS

To display the enriched sensor stream data with

semantic annotations, a web based IoT real-time

application is developed. The interface of the

software is shown in Figure 5, which enables real

time water quality monitoring through static and

mobile sensors. This software includes modules for

system administration (to manage users, user groups,

rights of users, and change password), enables

definition of the continuous queries, executes of the

ad-hoc queries, and configuration of the WSNs

metadata.

WQMS software executes continuous queries of

the proposed model to display information. The

displayed information in the textboxes for each

parameter is obtained from Processor Data Streams

through continuous queries. The information

displayed in the charts is obtained from Working Data

Streams, while the semantic annotations data that

indicates the water status is obtained from Working

Data Stream Annotations. As mentioned above,

Working Data Streams represent a fixed sliding

window which is configured in the WQMS by a

certain pre-configured size, say 15. This means that

the charts show the last 15 measurements for each

sensor. As soon as the data reaches the system from

the WSNs, the trigger for execution of queries

continuously is activated.

WEBIST 2020 - 16th International Conference on Web Information Systems and Technologies

Figure 2: 'ont-core.owl' management

model in WQMS.

Figure 3: System Architecture: implementation of proposed data stream.

Figure 4: Outputs of proposed model in WQMS.

A Management Model of Real-time Integrated Semantic Annotations to the Sensor Stream Data for the IoT

7 CONCLUSION AND FUTURE

WORK

In this paper, the proposed model of integrated

semantic annotations into the sensor stream data for

the Internet of Things is described.

The model supports managing stream data of

homogeneous sensors, real-time integration of

semantic annotations to the sensor stream data,

continuous queries on streaming data, ad-hoc queries,

outlier validation of streaming data, archive stream

data with semantic annotations for applications that

need to answer queries form archival store (persistent

data stored).

The model supports the following standards in

order to encode semantic annotations and data

observed by sensors: Sensor Web Enablement

(SWE), respectively version 2.0 of the Sensor

Observations Service (SOS) standard that relies on

the Open Geospatial Consortium (OGC), Observation

& Measurement (O&M).

To validate the proposed conceptual model, we

have developed a prototype for water quality

monitoring, named Water Quality Monitoring System

(WQMS). Applying advanced technologies of the

Internet of Things such as WSNs, the WQMS enables

water quality monitoring in real time.

Several extensions of the proposed model that can

be considered for the future are:

1. To advance annotation techniques, such as

XPath, for integration and interpretation of

the semantic annotations in real-time into

heterogeneous sensor observation data and

metadata with context in the Internet of

Things.

2. To advance the components Outlier Stream

Validator & Classificator of the proposed

model by implementing some advanced

outlier detection algorithms for real time

unsupervised anomaly detection.

3. To evaluate the system performance and to

compare the proposed model with other

existing similar management schemes.

REFERENCES

Bytyçi, E., Sejdiu, B., Avdiu, A., & Ahmedi, L. (2019). A

Semantic Sensor Web Architecture in the Internet of

Things. Semantic Web Science and Real-World

Applications (pp. 75-97). IGI Global.

Elnahrawy E. (2003). Research Directions in Sensor Data

Streams: Solutions and Challenges, Rutgers University,

Tech. Rep. DCIS-TR-527.

Lazarescu, M. T.. (2017). Wireless Sensor Networks for the

Internet of Things: barriers and synergies. Components

and Services for IoT Platforms. Springer.

Lin, S.Y., Li, J. B., and Yu, Ch. T. (2019). Dynamic Data

Driven-based Automatic Clustering. Sensors and

Materials, Vol. 31, No. 6 (2019) 1789–1801.

Rajaraman, A., Leskovec, J., and Ullman, J. D.. (2014)

Mining of Massive Datasets. Cambridge University

Press.

Sejdiu, B., Ismaili F., and Ahmedi L., (2020). Integration of

semantics into sensor data for the IoT - A Systematic

Literature Review. International Journal on Semantic

Web and Information Systems (IJSWIS). Volume 16,

Issue 4, Article 1.

Sejdiu, B., Ismaili F., and Ahmedi L., (2020). A real-time

integration of semantics into heterogeneous sensor

stream data with context in the Internet of Things. The

15th International Conference on Software

Technologies (ICSOFT 2020). July 07 - 09, 2020,

Lieusaint - Paris, France.

Shi, F., Li, Q., Zhu, T., Ning, H. (2018). A Survey of Data

Semantization in Internet of Things. Sensors 18(1).

Khan I., Jafrin R., Errounda F., Glitho R. (2015). A data

annotation architecture for semantic applications in

virtualized wireless sensor networks. In Integrated

Network Management, 2015 IFIP/IEEE International

Symposium.

Vera, D., Izquierdo, Á., Vercher, & J., Gómez, L., (2014).

A Ubiquitous Sensor Network Platform for Integrating

Smart Devices into the Semantic Sensor Web. Sensors

2014, 14, 10725-10752.

Pradilla, J., Palau C., & Esteve, M. (2016). SOSLITE:

Lightweight Sensor Observation Service (SOS) for the

Internet of Things (IOT). ITU Kaleidoscope: Trust in

the Information Society, Barcelona.

Wang, X.; Wei, H.; Chen, N.; He, X.; Tian, Z. (2020) An

Observational Process Ontology-Based Modeling

Approach for Water Quality Monitoring. Water, 12,

715.

Xiaomin, Zh., Jianjun, Y., Xiaoci, H., Shaoli, Ch. (2016).

An Ontology-based Knowledge Modelling Approach

for River Water Quality Monitoring and Assessment.

Procedia Computer Science, Vol. 96, Pages 335-344.

Yinbiao, S. and Lee, K.. (2014). Internet of Things:

Wireless Sensor Networks. International

Electrotechnical Commission (IEC), Switzerland.

Yu, K, Shi, W., & Santoro, N. (2020). Designing a

Streaming Algorithm for Outlier Detection in Data

Mining – An Incremental Approach. Sensors.

WEBIST 2020 - 16th International Conference on Web Information Systems and Technologies