TSHD: Topic Segmentation Based on Headings Detection (Case Study: Resumes)

Table 3

An illustrative example shows the output of each stage of the TSHD algorithm.

Raw data (a resume)	John Smith
	Software engineer with over 6 years of experience in applications developing for international companies
	Personal Data
	Address: 1234 McCoy Road phone: [phone]
	E-mail: [email protected], [email protected]
	Experience
	Employment at Maryland University
	Teaching assistant in the Department of Computer Science from 1/1/2017
	Educational Qualification
	Graduated with a Bachelor of Information Technology
	From University of Maryland 2011–2016
	Skills
	Programming skills:
	Python-java-C++
	Software skills:
	Linux-Windows
	Certifications and Awards
	International Computer Drivers License (ICDL) ACM ICPC 2019 gold medal
Preprocessing stage output for Th = 3 words	((0, “john smith”), (1, “software engineer”), (3, “personal data”), (6, “experience”), (9, “educational qualification”), (12, “skills”), (13, “programming skills”), (14, “python-java c++”), (15, “software skills”), (16, “Linux-Windows”), (17, “certifications and awards”))
First scan output	(“Personal info”: 3, “education”: 9)
Second scan output	(“Personal info”: 3, “experience”: 6, “education”: 9, “skills”: 12, “certifications”: 17)
Segmentation output (JSON)	(“personal_info”: “John Smith\nSoftware Engineer\nWith over 6 years of experience in applications developing for international companies.\nAddress: 1234 McCoy road phone: [phone]\nE-mail: [email protected] [email protected]”,; “experience”: “Employment at Maryland University\nTeaching Assistant in Department of Computer Science from 1/1/2017”,; “education”: “Graduated with Bachelor of Information Technology\nFrom University of Maryland 2011–2016”,; “Skills”: “Programming skills:\nPython-Java-C++\nSoftware Skills:\nLinux-Windows”,; “certifications”: “International Computer Drivers License (ICDL)\nACM ICPC 2019 gold medal\n”}