- Ontology / Semantic Web
- Ontology modeling: theory, experiment and application (The OWL Language)
- Concept modelling and reasoning – OpenMind and ConceptNet
- The Semantic Web Vision
- Ontology / Querying
- Semantic answers – meaningful answers to web-queries
- Ontology-based querying
- Searching / Intelligent Systems
- On use of Conceptual Knowledge for extraction and retrieval of text
- Ontologies in relational database systems
- Fuzzy Clustering
- Ontology building
- Game playing
Ontology modeling: theory, experiment and application
Description logic is an established formalism for ontology modeling and OWL is probably the most widespread description logic based modeling language. The aim here is to study the theory within and behind OWL and to perform an experimental modeling of a given domain using the Protégé-OWL tool. In addition potential applications of the derived ontology can be studied and/or develop.
Concept modelling and reasoning — OpenMind and ConceptNet
The Open Mind Commonsense project is an attempt to make computers smarter by providing millions of pieces of ordinary knowledge that constitute "common-sense", all those aspects of the world that we all understand so well we take them for granted. A main goal is to build a repository of commonsense knowledge by making it easy and fun for people all over the world to work together and provide pieces of commonsense.
ConceptNet is a knowledge base that is generated automatically from 700.000 sentences of commonsense.
The purpose of this project is to study the idea of commonsense repositories and to compare ConceptNet to other "knowledge bases" such as WordNet and Cyc.
Develop an application for instance as a tool that apply ConceptNet. Take inspiration from previously developed experimental application, as: ARIA (Liu & Lieberman, 2002) that observes a user writing an email and proactively suggests photos relevant to the user's story, or Overhear (Eagle et al., 2003) a speech-based conversation understanding system that uses common sense to jist the topics of casual conversations.
Ontology applications: The Semantic Web
The Semantic Web vision is about semantic annotation of documents and the general idea is that documents on the web should be accessible based on semantic content rather than on simple word occurrences and linking statistics. To provide such annotations knowledge resources for the semantic description reference are needed. These would typically be some kind of Ontologies.
While HTML is used for formatting of documents and basic XML allows for structuring of documents, RDF provides means for metadata and thereby for more detailed descriptions. Thus also for semantic descriptions.
The objective in a Semantic Web project can be to study the semantic web vision, as well as a variety of contemporary ontology modeling approaches, in more detail. In addition experimental validation of appropriate tools and languages can be performed on the basis of a selected domain.
Semantic answers — meaningful answers to web-queries
Develop an approach to "semantic clustering" of answers to web-queries.
A typical query to a search engine has thousands or hundreds of thousands of objects in the answer and the user has to rely on search engines ranking to show the "most interesting" of these as the first listed.
An alternative to an ordered list of objects as answer is to provide a list of topics. Such a list may be derived from a grouping of semantically meaningful groups — or clusters — that is, groups of objects that to some extend is on the same topic.
One approach in this direction is to provide to each object a semantic description, in the form of conceptual keywords, and to group the objects in the answer by statistically clustering based on these keywords.
The project may involve use of linguistic resources, such as Wordnet, and ontologies. Perspectives for semantic answers in the context of the semantic web vision can also be relevant.
Ontologies and applications: Ontology-based querying
The general idea in is to provide content-based information retrieval. Through text and context analysis concepts can be identifyied in text, extracted as descriptions and situated in an ontology. Based on ontology and descriptions queries can be given a flexible conceptual rather than a strict lexical interpretation. A ontology-based querying project can study extraction (indexing) as well as the query evaluation.
On use of Conceptual Knowledge for extraction and retrieval of text
The purpose of the project is to study on a theoretical as well as an emperical level (development of prototype for a selected base of texts) the use of conceptual knowledge in extraction and retrieval of text. Conceptual Knowledge can range from simple "controlled" lists of words over taxonomies of multi-word concepts to general ontologies, and may also include heuristic knowledge as for instance statistically based association structures over concepts (simple example: a word coocurrence matrix). Extraction using conceptual knowledge is about making descriptions of text fragments with reference to the conceptual knowledge (simple example: using only the controlled words) by means of for instance simple word recognition, mining, or Natural Language Processing. Retrieval may involve pre-compiled descriptions (simple example: crawled index for a web search engine) and involve the derivation of ranking functions based on (and respecting) the conceptual knowledge. The text base may be a small or larger collection of selected texts and crawling and search on the internet may also be taken into consideration (text base = all available web pages) The goal of the project should not be to cover all aspects within this broad area, but rather to narrow on selected problems within this area.
Ontologies in relational database systems
Develop an approach to represent and query an ontology in a relational database system.
In many areas, such as biology and medicine, ontologies are used to annotate objects of interest, such as biological samples, clinical pictures, or species in a standardized way. In these applications, an ontology is merely a structured vocabulary in the form of a tree or a directed acyclic graph of concepts. Typically, ontologies are stored together with the data they annotate in relational databases. Querying such annotations must obey the special semantics encoded in the structure of the ontology, i.e. relationships between terms, which is not possible using standard SQL alone.
The main objective in this project is to study approaches to represent an ontology in a relational database and to consider the challenges for efficient querying of this — taking into account for instance use of pre-computed indexes as alternative to recursion.
Clustering involves the task of dividing data points into homogeneous classes or clusters so that items in the same class are as similar as possible and items in different classes are as dissimilar as possible. Clustering can also be thought of as a form of data compression, where a large number of samples are converted into a small number of representative prototypes or clusters. Depending on the data and the application, different types of similarity measures may be used to identify classes, where the similarity measure controls how the clusters are formed. Some examples of values that can be used as similarity measures include distance, connectivity, and intensity.
In non-fuzzy or hard clustering, data is divided into crisp clusters, where each data point belongs to exactly one cluster. In fuzzy clustering, the data points can belong to more than one cluster, and associated with each of the points are membership grades which indicate the degree to which the data points belong to the different clusters.
The purpose of this project is to study clustering, to analyze the impact of fuzzyfication and to develop and implement a fuzzy approach for a chosen application.
Learning and game playing
Develop an agent for playing a selected game, for instance a board game, using an adaptive approach where the agent learns from experiences. Reinforcement learning is suggested as an approach that requires no prior knowledge or correct examples.