Speech recognition of named entities with word embeddings to display relationship information

An apparatus for speech recognition, the apparatus comprising:

• a microphone;
• a memory;
• and
• a processor,
• wherein the microphone recognizes speech uttered by a user, and
• wherein the processor is configured to:
• extract the speech uttered by the user to convert the speech into a text;
• recognize a named entity based on the text;
• structure a relationship of the named entity based on a word embedding method by:
• performing tokenization by separating the text into sentences as units and then identifying each word in a sentence among the sentences;
• and
• performing embedding by representing the identified word by a vector and vectorizing the word through part-of-speech tagging and a named entity class, wherein a similarity between features is measured through weight calculation between the features according to an unsupervised learning method, and the named entity with a high similarity for a threshold is clustered;
• determine an utterance intent based on the text corresponding to the speech of the user by:
• recognizing the named entity,
• recognizing a relationship information of the structured named entity, and
• restructuring a relationship of the named entity to reset the relationship information, when accuracy of the utterance intent is small compared to a preset target value, wherein a threshold in the restructuring of the relationship of the named entity is smaller than a threshold in the structuring of the relationship of the named entity previously performed;
• provide response information corresponding to a determination result of the utterance intent;
• display the relationship information of the structured named entity;
• determine a meaning of the named entity in consideration of the relationship information.