https://www.um.edu.mt/library/oar/handle/123456789/14673 2026-04-15T16:18:00Z 2026-04-15T16:18:00Z https://www.um.edu.mt/library/oar/handle/123456789/100948 2022-08-31T07:00:57Z 2012-01-01T00:00:00Z

Title: Fusing and recommending news reports using graph-based entity-relation representations Abstract: When an event occurs in the real world, news reports describing this event start to appear on news sites on the World Wide Web within a few minutes of the occurrence of that event. If the event is significant, numerous news reports will appear on different sites, and each report will give its own description of the event based on the sources of information available to its author. Moreover, as time passes, each news site may publish new reports related to that event that will contain information that has just been discovered. For a person to obtain all the details related to a particular event, he/she will have to read through all the reports covering that event. The multitude of news reports being published on a continuous basis on the World Wide Web also presents an issue of information overload on users. A user would need to sift through a huge number of news reports to identify those reports that are of interest to him/her. News aggregator web sites may cluster related news reports, but they do not attempt to fuse the reports into a single document that contains all of the pertinent information about a single event without any repetition. Such web sites also tend to display news reports chronologically, and a user who tracks an event over the course of several days must sift through them to identify previously unread material. Tracker news reports tend to repeat information that the user may have already read. Some news aggregator web sites and other web services will alert users to breaking news about types of events, or more typically, about news involving a named entity or event type (e.g. earthquake). However, the user must generally intervene to provide details of the entity or event type to track. In this thesis, we tackle a number of research problems: in theory, a user can identify any RSS feed as a source of news he/she would like to receive; we then cluster reports about related news received from the separate RSS feeds as they arrive; we fuse the reports into a single document, trying to preserve a logical order in which sub-events occur and eliminating repetition; new reports related to an existing cluster are integrated into the fused document; the user's interaction with a fused report is monitored in such a way that information that the user has already read is summarised so that in the next visit the user can focus on the new (novel) news; a user model is maintained to automatically identify entities and event types that appear to be of interest to the user so that he/she can be automatically alerted if a related new event occurs. We have developed the JNews news portal to implement our approach and to provide an evaluation platform to measure its ability to: i) cluster related news reports from disparate sources; ii) fuse related reports into a coherent document with minimal or no repetition but preserving all the information contained in the source reports; iii) provide an adaptive reading environment that automatically summarises information in reports that have already been read; iv) automatically identify entities and event types that the user is likely to be interested in based on their past interactions with JNews to make personalised recommendations about previously unread breaking news. As we do not know the number of clusters in advance, JNews uses a modified K-Means clustering algorithm. We represent information contained in news reports using a simplified version of Sowa's Conceptual Graphs. The graph representing a news event contains entities and their relationships. Information from related news reports is merged into a single graph. We keep track of the source sentences that express the relationships. The fused report is generated using the maximally expressive set of sentences, i.e. the sentences that contain most information about the entities and their relationships in the news report, and ensuring that all entities and relationships are expressed in the fused document. The advantage of using a simplified conceptual graph as the logical representation is that the entities and their relationships are represented canonically. We use the same graph to extract underlying patterns in information about types of events and/ or entities. If a user tends to read different fused reports about the same entity or event type then we can recommend similar breaking news to the user. In addition, we can recommend news using collaborative techniques. The user model is represented as a vector of weighted keywords. We use a summarisation technique, whereby the repetition of information across different documents is considered to be an indication of salience of that information, to present summaries of a fused report (containing only the most important information) that have already been read by a user. All components of JNews were designed to run fast without excessive computational resources so as to function well in an operational environment and be able to handle large amounts of data. The evaluation of JNews is performed on its three main components the Document Clustering Component, the Document Fusion Component, and the Information Filtering (recommendation) Component. The Document Clustering Component was evaluated using three different datasets. We found that our Document Clustering Component is very good in performing fine-grained clustering, but performs rather poorly when performing coarser-grained clustering. The Document Fusion component was evaluated using a set of news reports downloaded from MSNBC News that cite their sources, and also using human evaluation. We show that the Document Fusion component is able. to capture most of the information found across different source documents whilst maintaining readability. A corpus of news reports downloaded from Yahoo! News is used to evaluate the Information Filtering component. The results obtained are better than the baseline Rocchio algorithm without negative feedback. Description: PH.D.ARTIFICIAL INTELLIGENCE

2012-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/95517 2022-05-11T12:29:59Z 2012-01-01T00:00:00Z

Title: Markerless localisation and path planning for the visually impaired Abstract: A substantial number of people are affected in various ways by visual impairments, most of which have no effective cure. Visually impaired people are presented with a challenging task when navigating through unfamiliar areas, and must depend on additional aids, namely, guide dogs, and any assistance provided by sighted people, thus limiting their independence and privacy. The goal of this study is to exploit the advantages of robust hardware, portability and widespread use of mobile devices to provide a means of guidance to the visually impaired in unfamiliar areas. With the aid of ongoing research being done in the area of computer vision technology, the proposed system will aid the user in identifying the location of the whereabouts while also providing navigation commands to arrive at the desired destination. The prototype was implemented using the client-server architecture, where the server implements both the place recognition module and the path planning module while the client acts as a peripheral device. The evaluation carried out on the implemented prototype gave satisfying results with the place recognition module having a high percentage in both the recall rate and precision. Both the place recognition module and the path planning module returned results in a relatively short period of time. Description: B.SC.ICT(HONS)ARTIFICIAL INTELLIGENCE

2012-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/95508 2022-05-11T10:01:33Z 2012-01-01T00:00:00Z

Title: Cross document coreference resolution and disambiguation for named entities in user web history documents Abstract: At present, search engine technology does not measure relevance according to the information needs of the user, but rather to the query searched. This is not an ideal approach since different users use identical queries for different information needs. One of the reasons this may happen is because of ambiguity between named entities such as persons, organisations, locations, etc. This dissertation attempts to solve the problem from the client's side by using a baseline streaming cross document coreference resolution approach to discover and disambiguate named entities from the user's web history. Several orthographic and contextual similarity measures are used for this task, including tests involving dice score and topic features. Cosine similarity measure is then used to calculate the similarity between the named entity and the clusters. The final score dictates whether the named entity is to be merged into a cluster or to be formed into a new one. Queries submitted to the search engine are then expanded by using coreference from the most similar cluster to that query. In order to evaluate the system, the WePS-2007 testing corpus is used for relevancy and accuracy. Description: B.Sc. IT (Hons)(Melit.)

2012-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/95504 2022-05-11T09:08:51Z 2012-01-01T00:00:00Z

Title: On and off body sensor fusion for a 3d motion controller Abstract: There are a variety of f engineering applications that would benefit from the development of a 3D motion controller including fields such as gaming, entertainment, sport and education. Companies such as Nintendo, Microsoft and Sony have recently developed a number of motion gaming controllers. These use both on-device motion sensors and off-device sensors to determine the controllers' position. In this final year project, signal processing and communication techniques are used to develop 3D orientation and position determining algorithms which can be used to realise an accurate 3D motion controller. This is achieved by fusing together information from on and off body sensors, with an Android smartphone providing the on-body sensors and a pair of cameras providing the off-body sensors. To develop the 3D orientation determination algorithm 3-axis gyroscope, accelerometer and magnetometer data are fused together to give orientation data. A complementary filter [ 6] is designed and implemented to achieve this as opposed to the slower, yet more traditional, Kalman filter. This filters out drift caused by the gyroscope and noise from the accelerometer and magnetometer to give accurate and robust orientation data. A revised complementary filter is then proposed to extend the solution to full 360° rotations. In the 3D position determination algorithm off-body cameras are used in conjunction with the Hough Circle Transform [32] to locate a spherical marker attached to a Smartphone in video feeds. Mono and stereo video position location techniques are then used to determine the 3D position relative to one camera. Three scenarios are examined using a single camera, a pair of parallel oriented cameras and a pair of perpendicular cameras. In the final setup the cameras are calibrated such that one is the reference whilst the other is unconstrained. The x, y, z coordinates of the marker are found from the closest distance between skew vectors emanating from the two camera models, with sub-centimeter accuracy. An algorithm is finally presented which can be used as the bases of a future improvement to automate the calibration process. The two algorithms are finally combined together leading to a successful 3D motion controller. Description: B.SC.(HONS)COMPUTER ENG.

2012-01-01T00:00:00Z