We develop the time serial module-based deep learning algorithms to improve the performance of a collaborative filtering recommendation system. Also, adding the event and seasonal information to the inference procedure to broaden the scope of recommendation candidates and that makes a more comfortable environment for elderly care users.

可紀錄使用者過去活動並導入深度學習模型以建置符合時序之適應性推薦模式，並量化環境因素如節日或事件以實現友善之認知服務，使智慧型居家機器人在陪伴過程中能讓被照護者感覺更聰明及人性化。

1.The technological inventions include interacting with a user according to the user＇s dialog inputs to understand his semantic to detect intention and need.
2.The technological inventions include providing intelligently recommendation lists according to user’s preferences, time, causal relationship, and scenario.