Jin Nakazawa Lab

Likeyboard: Instant Lifelog Data Sharing Using Smartphone Keyboard Interface

Person in Charge of the Project : Rei IshikawaIn recent years, various life log data have been accumulated for meals, sleep, exercise, and so on. Though these accumulated life log data are used for reviewing the activities of individuals using applications and web APIs, it is rare that they are used on a daily basis, such as sharing among users. By making the accumulated life log data easily accessible on a daily basis, it is possible to create a new community using life log data. So in this research we propose a Likeyboard that can share the life log data that is accumulated on a daily basis in various formats (customizable stance), Likeyboard uses the keyboard interface of the smartphone, It is possible to access the life log data from the application.

VASHO: Haiku generation by deep language model

Person in Charge of the Project : Hirono KawashimaResearch topics on natural language processing fields such as translation and summarization have been actively conducted. In this study, we work on the haiku generation, which is one of sentence generation tasks. When we tackle this task and treat Japanese, we have to address three issues: 1) the way to decompose them such as the part-of-speech, since it is not delimited by spaces in sentences, 2) the expression of Japanese since the kind of character used is enormous, and 3) the semantic representation from reading between the lines. In order to solve these problems, we propose the language model VASHO and aim to generate haiku.

HealthyStadium: Meal Photo SNS with other evaluation-based Healthiness Competition for Improving Eating Habits

Person in Charge of the Project : Yusaku EigenObesity caused by deterioration of eating habits has become a social problem in recent years. Continuous eating habit improvement methods are needed, and many meal management methods using calorie calculation have been proposed. However, since calorie calculation requires grasping food name and quantity and inputting of such information, much effort is required for every meal, which hinders continued use. In order to solve this problem, this research aims to develop a social media service using friends' health evaluation of dietary photographs and competition principle and to develop a sustainable eating habit improvement.