Method for Suggesting Suitable Location-words for Demand Sentences

The ability to converse with human beings is essential if robots and humans are to communicate smoothly. Human responses are characterized by spontaneity, and conversational depth frequently veers between responses, suggestions, in-depth topic exploration, new topic introduction, and so on. Therefore, in this paper we propose a method of suggesting suitable location-words (words or short phrases representing locations) for demand sentences that can be applied to the technology responsible for ensuring smooth communication of humans and robots. Starting with a demand sentence, our proposed method creates a demand-concept from words that make up the demand sentence, after which a Web search is conducted to determine whether or not the obtained candidate location-words represent locations by using a thesaurus. Suitable location-words are then selected by calculating the DoA between location-word candidates and the demand-concept. In our evaluation of this concept, 86.7% of the location-words that were proposed by 60 demand sentences were determined to be correct.