Download presentation
Presentation is loading. Please wait.
Published byJeffery Reeves Modified over 9 years ago
1
Semantic Matching of Candidates’ profile with Job data from Linked-in Ting Xiao, Sarabpreet Dhillon Demo12/3/2013
2
Quick Review- Interim Info 0 Data Collection 0 Build Keyword Set 0 Build The Relationship Between the Job and Personal profile on the Ontology 0 Compute the matching score based on the linkage 0 Result
3
Architecture Linked-In Python API Alchemy Python API Jena Framework Web Page Json ttl
4
Data Collection 0 Python LinkedIn Library 0 https://github.com/ozgur/python-linkedin https://github.com/ozgur/python-linkedin 0 Profile.py 0 Developer’s own profile is easy to extract 0 Or use search API to find other person’s profile 0 Job_extraction.py 0 According to the level of the Linked-In account, the query gives different numbers of jobs’ information 0 The result is sorted by the timestamp of posting
5
Data Collection 0 Example of Profile(profile.json)
6
Data Collection 0 Example of Job Description(job_description.json)
7
Build Keyword Set 0 The skills Keyword Set (profile.json) 0 Good to use 0 The Job Description has no Keyword Set (job_description.json) 0 Need to analyse 0 AlchemyAPI 0 https://github.com/AlchemyAPI
8
Build Keyword Set 0 Generate The Keywords Set from job description (Job_skills.json) 0 Example:
9
Build The Relationship 0 Web-Karma 0 https://github.com/InformationIntegrationGroup/Web- Karma https://github.com/InformationIntegrationGroup/Web- Karma 0 Prepare the ontology 0 http://mayor2.dia.fi.upm.es/oeg- upm/index.php/en/ontologies/99-hrmontology http://mayor2.dia.fi.upm.es/oeg- upm/index.php/en/ontologies/99-hrmontology 0 Convert the.ttl format to.owl by using the “OWL Syntax Converter” developed by the University of Machester 0 http://mowl-power.cs.man.ac.uk:8080/converter/ http://mowl-power.cs.man.ac.uk:8080/converter/ 0 Preload into the Web-Karma
10
Build The Relationship 0 The HR ontology
11
Build The Relationship 0 Build Profile Linkage
12
Build The Relationship 0 Build Linkage for Job skills
13
Build The Relationship 0 Build Linkage for Job Information
14
Compute the Matching Score 0 Load the profile.rdf and job_skills.rdf to Jena 0 Do the SPARQL query, count the number of the items which are same. 0 Example Query
15
Query Example select ?name where { ?y ?id. filter (?id = '5831587').?y ?x.?x.?x ?z.{ ?x ?name filter (?name = 'C++')} union { ?x ?name filter (?name = 'OpenGL')} union { ?x ?name filter (?name = 'Photoshop')} union { ?x ?name filter (?name = 'Windows')} union { ?x ?name filter (?name = 'Visual Studio')} union { ?x ?name filter (?name = 'Linux')} union { ?x ?name filter (?name = 'UML')} union { ?x ?name filter (?name = 'JavaScript')} union { ?x ?name filter (?name = 'Cocos2d')} union { ?x ?name filter (?name = 'Objective-C')} union { ?x ?name filter (?name = 'C')} union { ?x ?name filter (?name = 'OS X')} union { ?x ?name filter (?name = 'Computer Vision')} union { ?x ?name filter (?name = 'iOS')} union { ?x ?name filter (?name = 'Machine Vision')} union { ?x ?name filter (?name = 'Cocoa Touch')} union { ?x ?name filter (?name = 'Hibernate')} union { ?x ?name filter (?name = 'Xcode')} union { ?x ?name filter (?name = 'Mac OS X')} union { ?x ?name filter (?name = 'LaTeX')} union { ?x ?name filter (?name = 'MySQL')} union { ?x ?name filter (?name = 'Java')}}
16
What’s New 0 No open-source job application provides job semantic matching 0 All the job hunting web services just provide the searching result according to the latest creation time of the job position. 0 How to convert Json to RDF format 0 We checked lots of converters however few of them provide this feature, but most of them can only convert semantic file format. 1.RDF Translator: http://ebusiness-unibw.org (NG)http://ebusiness-unibw.org 2.Karma:http://github.com/InformationIntegrationGroup/Web-Karma (Good)
17
Result 0 Demo 0 Welcome to use our job matching engine!
18
Question?
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.