Serendipitous book explorer using personalized associative dictionaries

This paper proposes a system called AssocSearcher that utilizes associative dictionaries as assistance for exploratory search for books. Users of the system can extend their horizons to explore books by getting subtopic or related words from the other people’s associative dictionaries. The associative dictionaries are created as substitutes for individual thoughts. We generated associative dictionaries of individual persons using Word2vec and GloVe from sentences read or written by the individuals. This allows each associative dictionary to respond to related words that are characteristics of each person with a particular topic or keyword. Experimental results obtained by searching books using multiple associative dictionaries showed that participants are more aware of the possibility of exploration. We also found a dictionary familiar to a particular participant improves comprehension.