Constructing Semantic Knowledge Model based on Children Dictionary
Masood Ghasemzadeh · 2010
Natural Language is a rich source of knowledge. Processing natural language in a way that computer can understand and draw linguistic inferences from it is the main research track in the Natural Language Processing (NLP) field. Semantic Knowledge Model (SKM) plays an important role in modeling and representing different knowledge sources. In this domain, a project is defined to address the previous problems in SKM and propose a new solution to construct a new one. The whole research project is based on a hypothesis: to have an ideal knowledge model, we need to construct our knowledge model based on the available knowledge source (e.g. dictionary definitions), from a very basic level such as Children dictionary and then integrate and extend it to a higher level e.g. Adult dictionary. This master thesis tried to evaluate part of this hypothesis, the ability to expand the definitions in an automated way and its potential to have the same mechanism at higher level. Expansion is used to expand the definition i.e. replacing every meaningful word in it by its own definition. The expansion mechanism helps to gather the sparse knowledge in dictionary. In this direction, “Core words” are defined. They constitute minimal set of words from which all other words can be defined by, in the dictionary. Besides what have been mentioned about core words, they help to control how far a definition can be expanded (otherwise it could be expanded unlimitedly). The problems which we faced during expanding the dictionary definitions are analyzed and our suggested solutions summarized carefully. The license of definitions in the Oxford First Dictionary has been bought by AI lab at EPFL. Due to the time limitations, only the verbs in the children dictionary which correspond to 429 head-words were analyzed and expanded. Chapter 1 provides some basic information about Natural Language Processing, our motivations to start such a project, review of the three similar projects and our goal in this research. Chapter 2 details knowledge sources which we considered. In this chapter, we also analyzed the structures of the dictionary definitions. Chapter 3 introduces expansion mechanism and the relevant concepts. In chapter 4, practical issues in the expansion of children dictionary definitions are reveled. In this perspective, all the problems encountered during expansion steps are listed, categorized and analyzed. Conclusion and the future work is stated in the last chapter. Keyword: semantic model, knowledge representation, expanding dictionary definitions, core words.