Eureka! Institutional Repository

This diploma thesis examines the implementation of the web-based search engine that receives text questions in "natural language", process them based on NLP techniques (Natural Language Processing) and then attempts to answer the question of the user. A mechanism was developed where it takes as input, in natural language texts and uses the Google NLP API to analyze the text syntactically, semantically, morphologically, emotionally and grammatically, sentimentally. It then extracts metadata for parts of Speech POS, sentiment magnitude, recognizes and extracts the entities contained in them. Then it uses the analysis and metadata generated to match the query to a predefined template and sends a query request to the search engine. The search engine that is implemented in Elasticsearch, searches its index, consisting data on movies and actors based on the IMDB dataset, and presents the results through a web interface. In this phase, the following search techniques are applied to predefined taxonomies, faceted search, fielded search, spelling correction