This thesis introduces a new approach to designing a Message Handling Assistant (MA). It presents a model of an MA and an intention extraction function for text messages, such as emails and Newsgroups articles. Based on a speech act theory and the belief-desire-intention (BDI) theory of rational agency, we define a generic MA. By interpreting intuitive descriptions of the desired behaviours of an MA using the BDI theory and speech act theory, we conjecture that intentions of messages alone provide enough information needed to capture user models and to reason how messages should be processed. To identify intentions of messages written in natural language, we develop a model of an intention extraction function that maps messages to intentions. This function is modelled in two steps. First, each sentence in a message is converted into a tuple (performative, proposition) using a dialogue act classifier. Second, the sender's intentions are formulated from the tuples using constraints for felicitous human communication. As an investigation of the use of machine learning technologies for designing the intention extraction function, four dialog act classifiers are implemented and evaluated on Newsgroups articles. The thesis also proposes a semantic communication framework, which integrates the agent and Internet technologies for automatic message composing and ontology exchange services.

A Computer-Assisted sentence Stress Teaching and Learning Environment (CASTLE) is proposed and developed, in order to help learners of English as a Second Language (ESL) to perceive and produce English stress correctly. Sentence stress plays an important role in English verbal communication. Incorrect stress may confuse listeners, and even break down a conversation. Stress is also challenging for ESL learners to master. It is neither easy for them to produce nor easy to perceive stress. Learners tend to transfer the stress patterns of their first language into English, which is not always appropriate. However, stress has been overlooked in English language teaching classes, due to the time limits of a class and teachers’ lack of confidence of teaching stress. Thus, CASTLE is intended to help ESL learners to use sentence stress correctly. There are three modules in CASTLE: an individualised speech learning material providing module, a perception assistance module and a production assistance module. Through conducting an investigation into which voice features (i.e. gender, pitch and speech rate) makes a teacher’s voice preferable for learners to imitate, we find that learners’ imitation preferences vary, according to many factors (e.g. English background and language proficiency). Thus, the speech material providing module of CASTLE can provide individualised speech material for different learners, based on their preferred voice features. In the perception assistance module of CASTLE, we propose a set of stress exaggeration methods that can automatically enlarge the differences between stressed and unstressed syllables in teachers’ voice. These stress exaggeration methods are implemented by the manipulation of different prosodic features (i.e. pitch, duration and intensity) of the teachers’ voice. Our experimental results show that all our proposed exaggeration methods could help ESL learners to perceive sentence stress more accurately. In the production assistance module of CASTLE, we propose a clapping-based sentence stress practice model that is intended to help ESL learners to be aware of the rhythm of English language. By analysing the limitation of conventional categorical representation of stress, we propose a fuzzy representation which is intended to better represent the subjective nature of stress. Based on the fuzzy representation of stress, we propose three feedback models in order to help the learners correct their stress errors. In addition to the development of CASTLE, we also propose an enhanced fuzzy linear regression model which can overcome the spreads increasing problem encountered by previous fuzzy linear regression models.

Safety of the airborne platforms rests heavily on the way they are maintained. This maintenance includes repairs and testing, to reduce platform down time. Maintenance is performed using generic and specific test equipment within the existing maintenance management system (MMS). This thesis reports the work undertaken to improve maintainability and availability of avionics systems using an intelligent decision support system (IDSS). In order to understand the shortcomings of the existing system, the prevalent practices and methodologies are researched. This research thesis reports the development and implementation of an IDSS and the significant improvements made by this IDSS by integrating autonomous and independent information sources by employing a multi-agent system (MAS). Data mining techniques and intelligence agents (IA) are employed to create an expert system. The developed IDSS successfully demonstrates its ability to integrate and collate the available information and convert into valuable knowledge. Using this knowledge, the IDSS is able to generate interpreted alerts, warnings and recommendations thereby reasonably improving platform maintainability and availability. All facets of integrated logistics support (ILS) are considered to create a holistic picture. As the system ages, the IDSS also matures to assist managers and maintainers in making informed decisions about the platform, the unit under test (UUT) and even the environment that supports the platform.