Skip to content

kcvparser performance

Objective:

As part of parsing resume uploaded by end users in HRMS application, KPAI team is going to develop NLP rule based application in GATE tool. For this work we need to have some manually annotated resumes (at least 3000 documents) to develop expected performance application.

Annotation Types

As per current requirements from HRMS team, we are expecting following annotation types should be done by manual annotators.

  1. Personal Information
  2. Education Details
  3. Work Experience
  4. Skill Set
  5. Certifications
  6. Others

1. Personal Information Section

Annotation Type Description Features
Name Name of the resume holder initials, first name, middle name, last name
DOB Date of birth of the resume holder NA
Gender Gender of the resume holder NA
EmailId Email Id of the resume holder NA
MobileNumber Mobile Number of the resume holder NA
Address Address of the resume holder addressline1, addressline2, street, city, state, country, postal code

2. Education Details Section

Annotation Type Description Features
Course Course name ex: IT, CSE etc.. NA
CourseLevel Graduation Type ex: B.Tech, Masters etc NA
EducationMode Course completion mode ex: Regular, Distance etc NA
StartDate Course Start Date NA
EndDate Course End Date NA
InstituteName Name of the Institute NA
University Name of the University NA
Percentage (GPA) Percentage/GPA obtained in the course NA
Description Text area under this heading from resume NA

3. Work Experience Section

Annotation Type Description Features
Employer Organization name NA
Designation Designation of the employee NA
StartDate Work experience start date NA
EndDate Work experience end date NA
Description Text area under this heading from resume NA

4. Skill Set Section

Annotation Type Description Features
SkillName Name of the employee skills NA
Description Text area under this heading from resume NA

5. Certifications Section

Annotation Type Description Features
CertificationName Names of the employee certifications NA

6. Other Annotation Types

Annotation Type Description Features
Heading Names of different resume headings Ex: Work Experience, Technical Skills etc NA

Note: At present those are planned at my end, please review and suggest if any other section required from resume

Results from kcvparser

After developing rules (by observing more than hundred resumes)for parsing the resume to get our desired annotations ,kcvparser has achieved an accuracy of 79 percentage (as of now) when compared to manually annotated resumes(sample of twenty resumes are taken for comparison)

Note

This analysis is done purely based on Indian resumes in specific to software related resumes

performance .png

we achieved an overall average precision of 87%

overall average recall of 83%

Finally Overall average fMeasure : 79%