The ACM Recommender Systems (RecSys) is a world class, international conference for offering the state-of-the-art research outcomes, systems and techniques in the field of recommender systems.
“Recommendation is a particular form of information filtering, that exploits past behaviors and user similarities to generate a list of information items that is personally tailored to an end-user’s preferences.”
Seeing that RecSys brings along top research groups , together with many of the world’s leading industries working on recommender systems, I found it a great combination of research and practice to dig the latest innovations, introduce exciting challenges, and progress our understanding of recommender systems.
I attended the recsys 2019 which was held in Copenhagen, Denmark together with more than 900 people that came from academia and industry and would like to share a brief summary of what I found interesting.
Finding relevant jobs from a very big overload of vacancies is challenging. That’s where recommender systems show up in order to smartly support users in their job hunting journey.
Typical recommender systems use information about a user to recommend relevant jobs but they often miss exploration and user control.
In this paper written by Francisco Gutierrez Et al., an interactive dashboard; “Labor Market Explorer” has been introduced to complement the existing career finder systems. This tool enables job seekers to investigate the labor market in a personalized way based on their skills and competences.
Labor market and the job descriptions are so dynamic which makes effective recommendations very challenging.
Explaining job recommendations to show competence match as well as supporting exploration and user control over broad and diverse recommendations.
To design the “Labor Market Explorer”, a user-centered methodology was applied to gradually improve the initial design. After every evaluation, feedback was addressed in the next design, which was then again evaluated. Figure 2 shows the studies and the participants in each iteration. The main goal of each study and the most important outcomes are summarized.
They designed the application following a client-server architecture — see figure below. The server includes:
- The MyCareer API contains information about the profile of the job seeker in XML format, consisting their desired jobs, education, and detailed competencies and skill related information that is entered by job seekers in the MyCareer system of the Public Employment Service VDAB.
- The ELISE1 component provides a service to match the profile of the user with open vacancies registered in the platform that was developed by VDAB, using a knowledge-based recommendation technique to suggest relevant vacancies to job seekers .
- The list of parameters (Figure 4) used by the knowledge-based recommender system of VDAB was presented and explained to the participants.
- participants were asked individually to rank them by personal importance/preference
3. Results has been collected (Figure 5) and analyzed (Figure 6). Based on the parameter ranking and input of the focus groups, they elaborated nine different designs that present information about jobs.
Design (a): presents the different parameters that were deemed important in four different clusters: location, work organization, competencies, and diploma. This design was selected as the main starting point for the first prototype.
Design (b): on the left side, the activities of a job seeker, in the middle, the vacancies and on the right side the distance are presented
Design (c): combines the idea of presenting activities with dots and stars, as well as the map for detailed location information.
4. Based on the feedback gathered from the previous iterations, following design goals for final prototype has been defined:
a. Exploration/control: job seekers should be able to control job recommendations and filter out the information flow coming from the recommender engine. (map and filtering component)
b. Explanations: recommendations and matching scores should be explained, and details should be provided on-demand. (vacancies table component)
c. Actionable insights: the interface should provide actionable insights to help job-seekers find new or more job recommendations from different perspectives.
Research questions and answers
1. Is enabling job seekers to use those visualization techniques helping them to find, understand and explore job recommendations?
. The approach is perceived as effective to explore job recommendations.
. Most participants felt confident and will use the explorer again.
. Explorations contribute to support user empowerment.
. A diverse set of actionable insights were mentioned by participant.
2. Do personal characteristics like, age or background influence the user understanding and user interaction with such an interface?
. The explorer was slightly better perceived by older participants (45+).
. participants in the technical group engaged more with all the different features of the dashboard.
. Non-native speakers, sales and construction groups engaged more with the map.
. The table overview was perceived as very useful by all user groups, the interaction may need further simplification for some users.
This paper has been discussed a user-centered design process involving job seekers and job mediators, enable job seekers to explore current vacancies and their required competencies, as well as how these competencies map to their profile. Evaluation results indicate the dashboard empowers job seekers to explore, understand, and find relevant vacancies, mostly independent of their background and age.
In the end I would like to thanks Arinti for the opportunity they gave me to attend Recsys 2019.
This blog was originally published on https://blog.arinti.be/a-summary-of-knowledge-based-job-ecommender-systems-2aa2fa46e56a
Disclaimer: This is not original work, the aim of this post is to spread and make more accessible the content of the paper “Explaining and exploring job recommendations: a user-driven approach for interacting with knowledge-based job recommender systems” by Francisco Gutiérrez and et al.