Text Labeling Applied in Shopping Assistant Robot Using Long Short-Term Memory

摘要

Shopping assistant robot plays a role as shopping guide in the supermarket. In this paper, we focus on early text labeling tasks which are crucial to design of robot conversation system and instruction selecting system. Part-of-speech tagging and named entity recognition are two fundamental tasks in nature language processing. In this paper, we need to implement both tasks in a robot by employing deep learning technics. The aim of this paper include: 1. Apply deep learning framework TensorFlow presented by Google to integrated robot development platform named Turtlebot2 which is provided by YUJIN ROBOT CO. 2. Construct text labeling model suitable for shopping assistant robot using a special recurrent neural network named long short-term memory.