| DMU logo

PhD Opportunities

There are a number of PhD opportunities, these include:

 

Cognitive Mobile Learning Platform Utilising Mobile Social Media Apps

First Supervisor: Dr. Aladdin Ayesh

Abstract: Mobile social media apps are the norm nowadays. For any business these apps are important to consider as part of the business plan. Considering the age groups using these apps it is equally important to consider as part of educational plans. Similarly, the wide spread of smartphones capable of delivering multi-modal interfaces enables the development of cognitive systems, e.g. that is capable of discerning emotions, at the same time increases user’s expectations from their devices and applications. This project will look at bringing these two areas together in developing the theoretical specifications of cognitive mobile learning platform and delivering a prototype demonstrating its practical use and applications.  

Back to top|

Large Scale Data Fusion for Sentiment Analysis from Multiple Social Media Sources

First Supervisor: Dr. Aladdin Ayesh

Abstract: The project will explore different techniques of data fusion and data mining to develop advanced processes for sentiment analysis over large-scale datasets. The collected data will be mostly unstructured or semi-structured textual data from multiple social media sources such as Facebook, twitter, news services RSS, etc.  Agent-oriented systems will be investigated to provide the distributed framework that enables dealing with multiple data sources. Once a distributed mechanism framework is established, it will be enhanced with data fusion techniques to collate, by means of classification and categorization, the data in relevant contexts. Graph-based techniques such as neural networks and cognitive maps, with minimalist dictionaries based text mining will be used in this process. These techniques have been used successfully in previous projects at DMU such as multi-lingual document classification by self-organizing maps, crime profiling by dictionary and local grammar based text mining. Each identified topic context will be mapped to sentiment context(s), which will guide the text mining process through the textual data. A graph-based map can then be constructed that can be then used in predicting sentiment. This concept was trialled already, in collaboration with Nuremberg Institute of Technology Georg Simon Ohm, where swarm intelligence techniques were used to construct a graph-based map to predict stock market sentiments from news feeds. This provides another dimension for an added innovation. Emotion models developed in previous projects at DMU will be used to develop the sentiment labelling and the underlying semantic definition within a scheme or labelling logic. This labelling logic will be applied to the output of text analysis giving sentiment snapshot of user interest as presented to the system by keywords or search zone. The full framework will provide the means of multi-label distributed classification to be developed, implemented and validated, which has many industrial uses especially in business intelligence and computational journalism.

Back to top|

Investigating the Inter-relationship between Emotion Detection and Personality from Computational Perspectives

First Supervisor: Dr. Aladdin Ayesh

Abstract: This project will look at the inter-relationship between emotion and personality starting from psychological studies and moving to computational models. The aim is to extend the capabilities of cognitive systems in emotion detection to support personalised emotion models and personalised user interfaces. In doing so, the project will investigate the personality and physiological influence of a given individual when feeling and expressing emotions. This will include the design of appropriate experimental settings using both inter-subject and intra- subject configurations to better understand the user impact on the detection process. This project is especially suitable for graduates of computer science joint degrees with psychology, education or sociology.

Back to top| 

A New Approach to Inferring Context Information from Sequential Sensor Data

First Supervisor: Dr Francois Siewe

Description: Ubiquitous computing (ubicomp for short) is ultimately the paradigm for next generation distributed systems where wirelessly networked computers disappear in the fabric of the user environment and interact calmly to provide the user with relevant information and services anywhere and anytime.  A typical ubicomp system monitors the user context (e.g. location, available resources and user activity) using a variety of sensors and uses that context information to decide autonomously what adaptation action to perform when this context changes, so as to minimise explicit user interactions. Thanks to the advances in technologies, the vision of ubicomp is increasingly becoming a reality with the proliferation of smart handheld devices such as smart phones and PDAs, equipped with a variety of sensors (e.g. GPS, accelerometers and gyroscope) to collect information upon the user context.  The applications of ubicomp are  found in healthcare, assisted living, intelligent transportation systems, and businesses, to name a few.

The quality of an ubicomp system is highly dependent on the quality of the context information obtained through sensors that are used to guide the behaviour of the system.  Wrong adaptation decision may be taken due to erroneous context information; what can be very dangerous in safety critical systems, e.g. in healthcare or assisted living system for disable people and the elderly. Unfortunately, sensors are never perfect; always produce noisy output and their accuracy is limited. Moreover, sensors do not often measure the desired context directly, but additional processing is required to infer the desired context information from the sensor data.

This research proposal aims at developing a new approach to inferring context information from sequential sensor data. The proposed approach will be applied to recognise user activities in a smart space, such as a smart assisted living system for disable people and the elderly. A prototype system will be developed and evaluated.

Back to top|

Probabilistic Calculus of Context-aware Ambients (pCCA)

First Supervisor: Dr Francois Siewe

Description: Ubiquitous computing (ubicomp for short) is ultimately the paradigm for next generation distributed systems where wirelessly networked computers disappear in the fabric of the user environment and interact calmly to provide the user with relevant information and services anywhere and anytime.  A typical ubicomp system monitors the user context (e.g. location, available resources and user activity) using a variety of sensors and uses that context information to decide autonomously what adaptation action to perform when this context changes, so as to minimise explicit user interactions. Thanks to the advances in technologies, the vision of ubicomp is increasingly becoming a reality with the proliferation of smart handheld devices such as smart phones and PDAs, equipped with a variety of sensors (e.g. GPS, accelerometers and gyroscope) to collect information upon the user context.  The applications of ubicomp are varied and include smart healthcare, smart assisted living, smart homes and smart transportation, to name a few.

Unlike traditional distributed systems, the design of ubicomp still faces a great deal of challenges due to that the users and portable devices are mobile,  communication is wireless mainly and the network topology is constantly changing in an unpredictable manner as nodes dynamically join and leave the system.  As a consequence, context information may be imprecise or incomplete at times.  This is critical for the behaviour of an ubicomp system as wrong adaptation decisions may be taken due to imprecision or incompleteness in context information.  There is a crucial need for a suitable framework for modelling and reasoning about the behavioural properties of ubicomp systems, prior to their actual implementation and deployment.

To address this problem, the Calculus of Context-aware Ambients (CCA) was developed in the Software Technology Research Laboratory (STRL) at De Montfort University.  It can be used to model and reason about mobility, context-awareness and concurrency. CCA has been applied to a number of ubicomp systems such as smart mobile learning, smart transportation, and smart communication protocol in vehicular ad-hoc network. However, it was found that CCA is limited in modelling and reasoning about imprecision, incompleteness and unpredictable behaviours; it lacks notations for quantifying these important features of ubicomp systems. Nevertheless, probabilistic models are widely used in the design and verification of complex system to quantify uncertainty and unpredictability.

This project aims at combining CCA and probabilistic models into a single formalism for modelling and reasoning about ubicomp systems. The resulting process calculus is called probabilistic CCA (pCCA for short). The research will build upon our previous work on CCA and (i) propose a syntax and a formal semantics for pCCA, with explicit notations for representing imprecision, incompleteness and unpredictable behaviours; (ii) develop techniques for reasoning upon the behaviour of pCCA processes, e.g. specific observational equivalence, bisimulation relations and type systems as customary in process calculi; (iii) develop toolkit to assist the analysis of systems; (iv) apply the proposed approach to a number of case studies such as the design and analysis of a smart assisted living system for disable people and the elderly; and a behaviour detection system for cars and unmanned vehicles (UMV) in a smart transportation system.

Back to top|

 

mobile-network2
 
Technology Degree Show
Technology Degree Show

The Technology Degree Show highlights the work and achievements of our final year students.

DMU research news target area image
DMU research news

Read about our latest research efforts.

De Montfort Open Research Archive (DORA) target area image
De Montfort Open Research Archive (DORA)

De Montfort University's research repository.