Đang chuẩn bị nút TẢI XUỐNG, xin hãy chờ
Tải xuống
In order to build a simulated robot that accepts instructions in unconstrained natural language, a corpus of 427 route instructions was collected from human subjects in the office navigation domain. The instructions were segmented by the steps in the actual route and labeled with the action taken in each step. This flat formulation reduced the problem to an IE/Segmentation task, to which we applied Conditional Random Fields. We compared the performance of CRFs with a set of hand-written rules. The result showed that CRFs perform better with a 73.7% success rate. . | Semantic Discourse Segmentation and Labeling for Route Instructions Nobuyuki Shimizu Department of Computer Science State University of New York at Albany Albany NY 12222 USA nobuyuki@shimi zu.name Abstract In order to build a simulated robot that accepts instructions in unconstrained natural language a corpus of 427 route instructions was collected from human subjects in the office navigation domain. The instructions were segmented by the steps in the actual route and labeled with the action taken in each step. This flat formulation reduced the problem to an IE Segmentation task to which we applied Conditional Random Fields. We compared the performance of CRFs with a set of hand-written rules. The result showed that CRFs perform better with a 73.7 success rate. 1 Introduction To have seamless interactions with computers advances in task-oriented deep semantic understanding are of utmost importance. The examples include tutoring dialogue systems and the one described in this paper a natural language interface to mobile robots. Compared to more typical text processing tasks on newspapers for which we attempt shallow understandings and broad coverage for these domains vocabulary is limited and very strong domain knowledge is available. Despite this deeper understanding of unrestricted natural language instructions poses a real challenge due to the incredibly rich structures and creative expressions that people use. For example Just head straight through the hallway ignoring the rooms to the left and right of you but while going straight your going to eventually see a room facing you which is north enter it. Head straight. continue straight past the first three doors until you hit a corner. On that corner there are two doors one straight ahead of you and one on the right. Turn right and enter the room to the right and stop within. These utterances are taken from an office navigation corpus collected from undergrad volunteers at SUNY Albany. There is a good deal of .
