الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 110 |  |  |
| | | from | 110 | | |
Abstract
The Internet of Things, which enables objects to connect with one another to exchange information and make choices, is a result of significant de- velopments in Internet and communication technologies. Several fields, including healthcare, agriculture, and manufacturing, have benefited from the development of the Internet of Things. One of the most crucial applica- tions of the Internet of Things is a system of monitoring daily life activities (DLA).
This monitoring system captures all potential sequential human activities associated with certain behaviour, such as behaviour of cooking, leaving the home, and going to sleep. As a result of this system, any missing or unusual activity related to a certain behaviour may be quickly noticed, alerting the person. In many circumstances, this warning might save a person’s life.
Sequential pattern mining is one of the most important data mining tasks that plays an important role in collecting and analyzing such a huge num-
vii
ber of possible daily life activities. Because it considers the right-ordered sequence of activities that a person follows in order to achieve a specific behaviour. For example, when analyzing the behaviour of the person to leave the home, we can discover that he/she undertakes the following se- quence of activities: ”putting on shoes, taking the mobile phone, taking the keys, and then opening the door.” As a result, if the system discovers that the person, for example, opened the door but did not take the keys, it notifies the person of the forgotten behaviour.
There have been a lot of studies done on mining sequential patterns, however, there are issues in those studies especially in DLA that make their usage challenging. These difficulties can be summarized as follows:
• There is an essential difference between the DLA datasets and tra- ditional datasets used by sequential mining algorithms. According to this issue, the algorithms have to be adapted to be able to deal with the new format of the DLA dataset.
• When applied to the DLA dataset, numerous important activity se- quences have been pruned based on the criteria used in sequential mining algorithms.
• Because of the significance of time for the system to quickly warn the person of the forgotten activity, the execution time of such algorithms requires some improvements.
On the other hand, there are other sequential mining-based studies spe-
ix
cially designed for the DLA dataset, in which the problem of pruning some interesting activity sequences has been overcome. However, because these studies employed classical algorithms like Apriori, these sequential patterns were discovered in extremely slow execution times.
The aim of this thesis is to develop an efficient sequential mining algo- rithm that overcomes the aforementioned difficulties. The proposed algo- rithm is called a Positional Representation-based Frequent Activities’ se- quence Mining algorithm PR − FASM ).
This thesis, which may be summed up as having five chapters, is orga- nized as follows:
Chapter 1 (Introduction): introduces a background of DLA, sequential mining algorithms, objectives, and our contributions.
Chapter 2 (Internet of Things and data mining challenges): studies the basic concepts of the Internet of Things and data mining challenges.
Chapter 3 (Problem statement and related work): formally introduces the problem statement and some of the basic definitions of sequential mining algorithms. It also introduces the related work of sequential pattern mining for analyzing daily life activities.
Chapter 4 (The proposed Positional Representation-based Frequent Ac-
tivities’ sequence Mining algorithm PR − FASM ) presents the proposed PR − FASM algorithm and some developed definitions, in addition to present and discuss experimental results that have been conducted on three real datasets, which showed the powerful and effectiveness of the proposed algorithm.
Chapter 5 (Conclusion and future work): summarizes the thesis and sug- gests some future work.