| Abstract [eng] |
As the amount of data stored, both online and in organizations, increases, the process of searching for information has become very difficult for the user. Traditional, keyword-based search engines cannot understand the semantics of the data presented and provide sufficiently accurate answers to user queries. Semantic technology and search are used to make the search more accurate. Semantic search is performed in ontologies, which define the concepts of certain subject areas, entities, their types, interrelationships, dependencies, and so on. Although there are different types of ontologies, this work focuses on applied ontologies for storage data in a specific subject area. The development of semantic search engines is also greatly influenced by the user interface, which must be not only functional, but also user-friendly and easy to understand for the system user. In this work, the main requirements of the natural language interface for ontologies were defined: adaptation to the Lithuanian language and preparation of the dictionary with as little manual work as possible. During this research project, the already developed natural language interfaces were examined and it was found that none of the analyzed user interfaces fulfilled all the requirements established during the research project. It has been established that the only SBVR based NKS is applicable to the Lithuanian language, but it requires additional dictionary preparation, which requires a lot of manual work. For this reason, there is a need to create a new solution that would allow the user to perform a semantic search in the ontology in Lithuanian. An analysis and specification of the requirements for a future natural language interface has formally described a method for analyzing questions and transforming them into queries. To test the method, a prototype of a natural language interface was realized, the operation of which was investigated in an experiment. The performed experiment helped to determine that the method allows to perform semantic search by asking questions in Lithuanian language. Although the accuracy of the answers is mediocre, it is possible to increase it by expanding the capabilities of the method. |
