2019Q3 Reports: Mentorship Co-Chairs
ACL 2019 is conducting two mentorship programs intended to provide support for newcomers to our conferences and/or for early-career researchers. These programs are a continuation of the mentoring programs started at NAACL 2019.
1. Sibling program: This program specifically targets newcomers who are attending a *CL conference for the first time. They are assigned a big sibling, who can take them around the conference, introduce them to their friends and colleagues and help them find suitable affinity groups. The main aim of this program is to ensure that first-time attendees are able to mingle and do not feel left out.
2. One-on-one mentoring: This program is focused on early-career researchers. They are assigned a mentor, usually an established researcher in the field, who can give them career advice such as choosing research problems to work on, managing time, and building a research identity. This program aims to ensure that young researchers get an opportunity to interact with senior researchers whose career paths they wish to emulate.
We are also co-ordinating with the Student Research Workshop (SRW) to help SRW attendees receive one-on-one mentorship.
To organize these programs, we sent out a survey to all registered participants of ACL and SRW (n=2901) on July 13th. This survey solicited their interest in participating in the program (as a mentor or a mentee) as well as some additional information such as affiliation, seniority, research interests, and so on, which would be used to make matches. As of July 16, we had 260 responses. Some statistics from the responses are reported below:
1. Number of big siblings: 27
2. Number of mentors: 42
3. Number of little siblings: 168
4. Number of mentees: 179
135 people have signed up to be both little siblings and mentees. By the sign-up deadline (July 18th), 388 people had responded. We are currently unable to access these because of a bug in Microsoft Office that has blocked access to form responses for everyone, but we hope it will get resolved soon.
Matching process: Similar to NAACL, we have a matching script to automatically create sibling and mentor-mentee matches. This script uses integer linear programming to create optimal matches while satisfying several constraints. The objective function used by the script is the sum of jaccard similarity scores between research interests of each mentor-mentee pair. Since sibling matches are inherently random, they are not included in the objective. The script tries to maximize this score (pairing mentors and mentees who have many matching interests). This maximization is subject to constraints such as not exceeding limits specified by mentors and big siblings (eg: if somebody asked for only one mentee, do not assign them more), ensuring that matched pairs are from different affiliations (to encourage more socialization) and constraining mentors to be more senior than mentees. We collected metadata such as seniority level and research interests from participants during the survey phase.
While the mentorship program at ACL 2019 is still ongoing, and we do not have a final set of lessons learned, this is what we have learned so far:
- The *CL community has a very large appetite for mentorship, possibly due to a sharp increase in the membership of our community over the past few years, which has resulted in a large number of new/early-career researchers. The immediate effect on the mentorship program, observed both during ACL 2019 and NAACL 2019, is a large gap between the number of people who are interested in receiving mentorship and those able to offer it. Future offerings of such programs will have to consider various solutions for growing the number of mentors.
- The forms used so far to collect mentor/mentee interest have posed issues. Google forms are not accessible in certain parts of the world, thus cannot be used. Microsoft Office forms have had issues, with the information collected from participants becoming inaccessible to the form organizers (a bug we are currently experiencing). If mentorship becomes an integral part of our conferences, as we hope it will, it might be worthwhile developing our own forms to collect mentorship interest.