Resume Classification and Dialogue State Tracking
Date: 2023-10-27 / 3:30 ~ 4:30 PM
Location: MSC E306
In recent months, our research endeavors at EmoryNLP Lab have steered towards innovative applications of natural language processing, encompassing both recruitment automation and the exploration of dialogue systems. This presentation will provide insights into two key projects I have been involved in.
- Resume Classification & Job Matching: The labor market has always been a dynamic domain, where the perfect alignment of candidate skills with job requirements can drastically improve recruitment outcomes. To this end, we embarked on a project to automate the classification of resumes based on their competence level. The developed model not only categorizes resumes but also bridges them to pertinent job descriptions. Through this initiative, we aimed to boost the efficiency of recruitment processes and provide a more tailored approach to job matching.
- Language Models in Dialogue State Tracking and Generation: The advent of large language models has reshaped the way we perceive and interact with machine-generated text. One significant arena where they show potential is in dialogue systems. Our research delved into gauging the efficacy of such models in the tasks of Dialogue State Tracking (DST) and Generation. DST, which involves comprehending the context of an ongoing conversation, and Generation, the production of human-like responses, stand as key components of an effective chatbot. Our exploration sought to answer the question: How well can contemporary language models fare in these tasks?
This presentation will shed light on the methodologies employed, the challenges faced, and the results attained in both projects.