Liang Qiu

Title: Innovations by Liang Qiu in Task-Oriented Dialogues

Introduction

Liang Qiu is an accomplished inventor based in San Jose, California. He has made significant contributions to the field of dialogue systems, particularly in the area of unsupervised structure extraction in task-oriented dialogues. His innovative approach has the potential to enhance the efficiency and effectiveness of dialogue systems.

Latest Patents

Liang Qiu holds a patent for "Systems and methods for unsupervised structure extraction in task-oriented dialogues." This patent proposes a novel approach that utilizes a Slot Boundary Detection (SBD) module. The SBD module tags utterances from training domains using the conventional BIO schema, but without the slot names. A transformer-based classifier is then trained to detect the boundaries of potential slot tokens in the test domain. The methodology assumes that while the state number is usually unknown, the slot number is given when analyzing a dialogue system. Detected tokens are clustered into groups based on the number of slots. The dialogue state is represented with a vector that records the modification times of each slot. Additionally, slot values are tracked through each dialogue session, labeling utterances with their corresponding dialogue states. The semantic structure is depicted by computing the transition frequencies among unique states.

Career Highlights

Liang Qiu is currently employed at Salesforce, Inc., where he continues to develop innovative solutions in dialogue systems. His work has garnered attention for its practical applications and contributions to the field.

Collaborations

Liang has collaborated with notable colleagues, including Chien-Sheng Wu and Wenhao Liu. Their combined expertise has further advanced the research and development of dialogue systems.

Conclusion

Liang Qiu's innovative work in unsupervised structure extraction for task-oriented dialogues showcases his significant contributions to the field. His patent reflects a deep understanding of dialogue systems and their complexities.