https://www.um.edu.mt/library/oar/handle/123456789/10780 2026-06-02T00:13:19Z 2026-06-02T00:13:19Z https://www.um.edu.mt/library/oar/handle/123456789/91755 2022-03-18T09:12:00Z 2015-01-01T00:00:00Z

Title: An emotional sensitive BCI in use with an expert system Abstract: Artificial Intelligence is always striving to create a human like system. It needs to learn, talk and react to the circumstances as we humans do. To develop something like this the system has often been added some senses, like vision or hearing, so that it can react to the world around it in an intelligent way. When we humans talk, we are sending non-verbal messages that indicate our mood and how much we're involving ourselves in the conversation at hand. This is something that the system will not get from the text and therefore it needs a new parameter to help the system perceive the meaning beyond the words. This parameter can be the brainwaves that are a very good indication on the mental state of the user thus helping the system to choose the right text to guide the user to a more emotionally connected state of mind. The aim of this project is to develop a system that is more responsive to the user, a system that adapts itself to make the users feel more emotionally involved in the conversation. The adaptability feature makes the user feel that the system is responding appropriately to him like a human-to-human session. So this will create a possibility for the simulation of human endeavors. The system will try to simulate a person that can tell personal details about the user with some psychic power. This simulation has been chosen so that an emotional state of mind is more reachable. (Ol)The aim will be achieved by developing an expe1i system that will incorporate rules involving the two inputs, text and brainwaves, in order to guide the user to an emotional state. This objective is broken down into two parts. (01.1) First, use brainwaves to guide the user to an emotional state, meaning that the preferred wave is Gamma (emotional), second is Beta (concentrated), third being Alpha (awake but relaxed) and finally Theta (bored). (01.2) Secondly use keywords in conjunction with brainwaves to guide the user to an emotional state.The system was tested using 12 subjects that were divided into 3 groups. Group A was doing the session with the keywords function deactivated. Group B was doing the session with the keywords function activated. Group C, the control group, was doing the session without being effected by the brain waves and the keywords, and always showed the statements associated with the gamma wave. The data that was collected from these groups was analyzed in two comparisons. The comparison of Groups A and C was done to evaluate the success of objective 01.1 and the comparison of Groups A and B was done to evaluate the success of objective 01.2. The comparisons were done on the bases of the occurrences of the different waves for the different groups and the waves as the subjects went through the session. Testing showed that Group A did better than Group C, and Group B did better than Group A. Description: B.SC.ICT(HONS)ARTIFICIAL INTELLIGENCE

2015-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/78261 2021-07-13T10:52:13Z 2015-01-01T00:00:00Z

Title: An online collaborative platform for the development of empirical grammars Abstract: Modern science often requires the collaboration of large groups of people and the sharing of specialized data and knowledge. Computational linguistics is not an exception, with projects often involving researchers and data from different parts of the world and different languages. In this document, a tool for internet collaboration in computational linguistics and its development are described. The tool is an online environment that enables users to manage and exploit different types of linguistic resources in a collaborative way. It features real-time editing of documents, visualization of complex objects and data pipelines for numerical and statistical analysis. An extensive HPSG interpreter has been developed at the same time, and is also documented here. It has been embedded in the online tool, and a visual feature structure editor has been included. With their aid, a small grammar of Spanish has been developed, as showcase of the system capabilities. One of these capabilities is the ability to extend the tool with arbitrary user code. In this document an example of this is provided, showing how the system might be used for interlingua or semantic translation. Description: M.SC.ARTIFICIAL INTELLIGENCE

2015-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/10988 2018-06-25T09:32:40Z 2015-01-01T00:00:00Z

Title: Sentiment analysis in Maltese Abstract: In today's modern age, the use of the web and social media sites for communication is on the rise, with more and more people each day making use of such facilities to express their opinion about one thing or another. This phenomenon has spread rapidly throughout the Maltese islands over the past few years, particularly on social media websites such as Facebook or online gazettes, where hundreds of reviews are posted on a daily basis by citizens wishing to make their voice heard about various subjects. Sentiment analysis refers to the task of analysing such reviews and classifying them as positive, negative or neutral, according to the overall sentiment of the opinion being expressed. In this FYP, we present a novel system capable of performing such a task for text written in Maltese. We propose a supervised machine learning, context based approach by which we aim to not only determine the optimal algorithm and parameters for achieving the best results possible with our system, but also to surpass a baseline accuracy of 34% obtained by a random classi er and reach that of 64% obtained through manually designed rules. Our system consists of two main components both capable of performing preprocessing, feature extraction and classi cation of text written in Maltese at a context window level, yet while one follows the more traditional machine learning approach where features are manually hand-crafted and passed on to classi cation algorithms, the other performs unsupervised feature extraction and makes use of deep learning classi ers to categorize the text. Through experimentation we determined that a Random Forest classi er in conjunction with 80% of our dataset for training and a 4 word context window was the optimal scenario to achieve the best results, and were successful in not only surpassing the baseline accuracy but also achieving a 62.3% accuracy through the use of the aforementioned classi er and parameters. Description: B.SC.IT(HONS)

2015-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/10987 2016-09-28T09:21:13Z 2015-01-01T00:00:00Z

Title: What is happening now? entertainment Abstract: Everyday people exchange data back and forth over the internet, describing their ongoing activities, their plans for the weekend and anything that comes to mind. Most data on the internet consists of unstructured text containing links to more structured information. To our knowledge, there is no automatic solution that identifies and extract entertainment events from multiple sources. Hence, the problem is finding a way we can use this data to extract useful information such as events.[1] In this research, we propose a solution for the automated detection and extraction of entertainment events, and the representation of these features in a structured format. The method proposed is divided in a number of steps: firstly, through supervised event classification the system finds out whether the data retrieved from various RSS feeds (e.g local press news sites) is an entertainment event or not. Secondly, the system annotates the documents that are classified to be entertainment events using different NERs included in GATE[2] pipelines to extract named entities (such as Event Date, Location, Participants and Organisations Involved). Moreover, we eliminate ambiguous dates and solve temporal expressions on these extracted event details. Furthermore, this event data is compared to previously extracted events and information aggregation is performed. Information aggregation is the coalescing of event data from multiple news reports detected to be referring to the same event. Finally, these event details need to be represented. For extensibility of the system the use of RDF model has been employed to represent these events in a semantic way. We even showcased an RDF API that allows others to perform free text searches on the entertainment events extracted by the system. In addition to this, we stored these entertainment events in database tuples to utilise such information for the web interface (front end UI). In this web interface we provide a way to search for entertainment events by their details using our simple RESTful API. In this prototype, we mainly focused on the detection and extraction of entertainment events in Malta. Nevertheless, such a system can be extended to retrieve and extract entertainment events for other countries. The results obtained based on our evaluation were very promising. Description: B.SC.IT(HONS)

2015-01-01T00:00:00Z