Title2Vec: a contextual job title embedding for occupational named entity recognition and other applications

• Published in: Journal of big data 2022-12, Vol.9 (1), p.1-16, Article 99
• Main Authors: Liu, Junhua, Ng, Yung Chuen, Gui, Zitong, Singhal, Trisha, Blessing, Lucienne T. M., Wood, Kristin L., Lim, Kwan Hui
• Format: Article
• Language: English
• Subjects: Big Data; Communications Engineering; Computational Science and Engineering; Computer Science; Conditional random fields; Data analysis; Data mining; Data Mining and Knowledge Discovery; Database Management; Datasets; Digital audio players; Embedding; Information Storage and Retrieval; Job titles; Machine learning; Mathematical Applications in Computer Science; Named entity recognition; Networks; Occupational mining; Recognition; Social computing; Social networks; Upstream; Word embedding
• ISSN: 2196-1115; 2196-1115
• DOI: 10.1186/s40537-022-00649-5

Occupational data mining and analysis is an important task in understanding today’s industry and job market. Various machine learning techniques are proposed and gradually deployed to improve companies’ operations for upstream tasks, such as employee churn prediction, career trajectory modelling and automated interview. Job titles analysis and embedding, as the fundamental building blocks, are crucial upstream tasks to address these occupational data mining and analysis problems. A relevant occupational job title dataset is required to accomplish these tasks and towards that effort, we present the Industrial and Professional Occupations Dataset (IPOD). The IPOD dataset contains over 475,073 job titles based on 192,295 user profiles from a major professional networking site. To further facilitate these applications of occupational data mining and analysis, we propose Title2vec , a contextual job title vector representation using a bidirectional Language Model approach. To demonstrate the effectiveness of Title2vec , we also define an occupational Named Entity Recognition (NER) task and proposed two methods based on Conditional Random Fields (CRF) and bidirectional Long Short-Term Memory with CRF (LSTM-CRF). Using a large occupational job title dataset, experimental results show that both CRF and LSTM-CRF outperform human and baselines in both exact-match accuracy and F1 scores. The dataset and pre-trained embeddings have been made publicly available at https://www.github.com/junhua/ipod .