De-identification of Privacy-related Entities in Job Postings

Kristian Nørgaard Jensen; Mike Zhang; Barbara Plank

De-identification of Privacy-related Entities in Job Postings

Kristian Nørgaard Jensen
, Mike Zhang
, Barbara Plank

NLPnorth

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

Abstract

De-identification is the task of detecting privacy-related entities in text, such as person names, emails and contact data. It has been well-studied within the medical domain. The need for de-identification technology is increasing, as privacy-preserving data handling is in high demand in many domains. In this paper, we focus on job postings. We present JobStack, a new corpus for de-identification of personal data in job vacancies on Stackoverflow. We introduce baselines, comparing Long-Short Term Memory (LSTM) and Transformer models. To improve upon these baselines, we experiment with contextualized embeddings and distantly related auxiliary data via multi-task learning. Our results show that auxiliary data improves de-identification performance.

Original language	English
Title of host publication	Proceedings of the 23rd Nordic Conference on Computational Linguistics
Publisher	Association for Computational Linguistics
Publication date	21 May 2021
Pages	210-221
Publication status	Published - 21 May 2021
Event	Nordic Conference on Computational Linguistics - Rejkjavik, Iceland Duration: 31 May 2021 → 2 Jun 2021 Conference number: 23

Conference

Conference	Nordic Conference on Computational Linguistics
Number	23
Country/Territory	Iceland
City	Rejkjavik
Period	31/05/2021 → 02/06/2021

Series	Linköping Electronic Conference Proceedings
Number	21
Volume	178

Keywords

De-identification
Privacy-related entities
Medical domain
Privacy-preserving data
JobStack corpus
Job postings
Personal data
Baselines
Long-Short Term Memory (LSTM)
Transformer models

Access to Document

JobStack_NoDaLiDa_2021Accepted author manuscript, 205 KB

Cite this

@inproceedings{4e400a886fcb4ba1ae10675b004fd96f,

title = "De-identification of Privacy-related Entities in Job Postings",

abstract = "De-identification is the task of detecting privacy-related entities in text, such as person names, emails and contact data. It has been well-studied within the medical domain. The need for de-identification technology is increasing, as privacy-preserving data handling is in high demand in many domains. In this paper, we focus on job postings. We present JobStack, a new corpus for de-identification of personal data in job vacancies on Stackoverflow. We introduce baselines, comparing Long-Short Term Memory (LSTM) and Transformer models. To improve upon these baselines, we experiment with contextualized embeddings and distantly related auxiliary data via multi-task learning. Our results show that auxiliary data improves de-identification performance.",

keywords = "De-identification, Privacy-related entities, Medical domain, Privacy-preserving data, JobStack corpus, Job postings, Personal data, Baselines, Long-Short Term Memory (LSTM), Transformer models, De-identification, Privacy-related entities, Medical domain, Privacy-preserving data, JobStack corpus, Job postings, Personal data, Baselines, Long-Short Term Memory (LSTM), Transformer models",

author = "Jensen, \{Kristian N{\o}rgaard\} and Mike Zhang and Barbara Plank",

year = "2021",

month = may,

day = "21",

language = "English",

series = "Link{\"o}ping Electronic Conference Proceedings",

publisher = "Association for Computational Linguistics",

number = "21",

pages = "210--221",

booktitle = "Proceedings of the 23rd Nordic Conference on Computational Linguistics",

address = "United States",

note = "Nordic Conference on Computational Linguistics , NoDaLiDa 2021 ; Conference date: 31-05-2021 Through 02-06-2021",

}

Jensen, KN, Zhang, M & Plank, B 2021, De-identification of Privacy-related Entities in Job Postings. in Proceedings of the 23rd Nordic Conference on Computational Linguistics. Association for Computational Linguistics, Linköping Electronic Conference Proceedings, no. 21, vol. 178, pp. 210-221, Nordic Conference on Computational Linguistics , Rejkjavik, Iceland, 31/05/2021.

De-identification of Privacy-related Entities in Job Postings. / Jensen, Kristian Nørgaard; Zhang, Mike; Plank, Barbara.
Proceedings of the 23rd Nordic Conference on Computational Linguistics. Association for Computational Linguistics, 2021. p. 210-221 (Linköping Electronic Conference Proceedings; No. 21, Vol. 178).

Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review

TY - GEN

T1 - De-identification of Privacy-related Entities in Job Postings

AU - Jensen, Kristian Nørgaard

AU - Zhang, Mike

AU - Plank, Barbara

N1 - Conference code: 23

PY - 2021/5/21

Y1 - 2021/5/21

N2 - De-identification is the task of detecting privacy-related entities in text, such as person names, emails and contact data. It has been well-studied within the medical domain. The need for de-identification technology is increasing, as privacy-preserving data handling is in high demand in many domains. In this paper, we focus on job postings. We present JobStack, a new corpus for de-identification of personal data in job vacancies on Stackoverflow. We introduce baselines, comparing Long-Short Term Memory (LSTM) and Transformer models. To improve upon these baselines, we experiment with contextualized embeddings and distantly related auxiliary data via multi-task learning. Our results show that auxiliary data improves de-identification performance.

AB - De-identification is the task of detecting privacy-related entities in text, such as person names, emails and contact data. It has been well-studied within the medical domain. The need for de-identification technology is increasing, as privacy-preserving data handling is in high demand in many domains. In this paper, we focus on job postings. We present JobStack, a new corpus for de-identification of personal data in job vacancies on Stackoverflow. We introduce baselines, comparing Long-Short Term Memory (LSTM) and Transformer models. To improve upon these baselines, we experiment with contextualized embeddings and distantly related auxiliary data via multi-task learning. Our results show that auxiliary data improves de-identification performance.

KW - De-identification

KW - Privacy-related entities

KW - Medical domain

KW - Privacy-preserving data

KW - JobStack corpus

KW - Job postings

KW - Personal data

KW - Baselines

KW - Long-Short Term Memory (LSTM)

KW - Transformer models

KW - De-identification

KW - Privacy-related entities

KW - Medical domain

KW - Privacy-preserving data

KW - JobStack corpus

KW - Job postings

KW - Personal data

KW - Baselines

KW - Long-Short Term Memory (LSTM)

KW - Transformer models

M3 - Article in proceedings

T3 - Linköping Electronic Conference Proceedings

SP - 210

EP - 221

BT - Proceedings of the 23rd Nordic Conference on Computational Linguistics

PB - Association for Computational Linguistics

T2 - Nordic Conference on Computational Linguistics

Y2 - 31 May 2021 through 2 June 2021

ER -

De-identification of Privacy-related Entities in Job Postings

Abstract

Conference

Keywords

Access to Document

Fingerprint

Cite this