RGCL Anniversary Highlights, Day 25
Published on Sep, 9 2022 by RGCL.
Interview with Prof Ruslan Mitkov
In this final instalment of the 25-year anniversary highlights series, we hear from the Director of RIILP and Head of RGCL, Prof Ruslan Mitkov.
Q: Prof Mitkov, you have been Director of the Research Institute and Head of the Research Group for the past 25 years. What led to you establishing these centres at the University?
I had the idea to establish a strong, internationally known research group from the moment I joined the University of Wolverhampton and this was the mission I was pursuing from the beginning. My mission was more than successful. The group I built went from strength to strength and today we can call it ‘internationally leading’.
In 2005 the University recognised that the two most successful research groups of the University were RGCL and SRCG (Statistical and Cybermetrics Research Group), and the then Dean of Research (extremely supportive, I wish she were still at the University) Prof Jean Gilkison proposed the establishment of RIILP on the basis of these research groups.
Q: We have heard comments from collaborators in the past that RGCL has “put the University of Wolverhampton on the map”. What made you choose Wolverhampton?
In 1995 had a job offer in Canada but upon reflection I decided to look for a job in the UK to be closer to my family. Wolverhampton was the first university I applied for and was offered the position.
Indeed, RGCL have put the University on the map. Many people know the University of Wolverhampton because of us and I wonder if the current University management is aware of that fact.
See the rubric ‘what others say about us’ at https://rgcl.wlv.ac.uk/. I would like to single out the following quotes:
Robert Galbraith, then Director of Innovation at National Board of Medical Examiners (NBME), USA:
“Wolverhampton is well-known for 2 things: the industrial revolution and … the Research Group in Computational Linguistics… They simply do magical things…”
Prof Will Lewis, University of Washington, former Principal Technical Program Manager at Microsoft:
“The Research Group in Computational Linguistics at the university of Wolverhampton is one of the premier CompLing programs in the world and literally put Wolverhampton on the map.”
Prof Alfonso Ureña López, President of the Spanish Association for Natural Language Processing:
“It is difficult to find many names in the field Natural Language Processing better known in the world than Prof Ruslan Mitkov and it is equally difficult to find many research groups in the field more successful than his research group.”
Prof Mitkov with the Rector of the University of Jaén and Prof Alfonso Ureña López
Q: How did you go about establishing the Group? Did you have some key collaborators at that time?
My strategy was to recruit outstanding PhD students to start with. I knew that those students who pay the fees themselves would normally choose Oxford or Cambridge. So my idea was to offer studentships to the best students from all over the world – we used a combination of overheads from successful projects and university funding allocated to us to offer bursaries to top PhD students. This is how I attracted outstanding students as Constantin, Catalina, An, Iustin, Wilker, and many others.
Q: What would you say are your most important achievements during your past 25 years at the University?
My achievements are many and it would be difficult to summarise them in this interview but creating RGCL is a top achievement itself. Many universities in the UK would do anything they can to have such a research group.
The majority of RGCL members have been outstanding professionals, competent, helpful and loyal. I would like to thank them from the bottom of my heart.
Q: Have there ever been times you thought about leaving or possibly moving the Group?
I was headhunted on many occasions by UK and overseas universities in the last 20 years but preferred to stay at the university. Because of the group. I felt I owed this to the group morally. And also because of the support I was enjoying from this university. Until about 2 years ago. Unfortunately, I cannot say the university has been very supportive in the last 2 years….
Q: You have attended so many international conferences and networking events over the years as an ambassador of the Group. Which ones have been your favourites?
Perhaps RANLP [Recent Advances in Natural Language Processing]… The good thing is that RGCL has always been one of its co-organisers…
Pictured: Prof Mitkov and collaborators in the early days of RANLP
Q: What would you say you are best known for in the research community?
I am perhaps best known for my contributions to anaphora resolution and the automatic generation of multiple choice tests. In fact, if there is a group considered to be ‘world leader’ in anaphora resolution, this is RGCL.
How did it start? In 1994 I had a paper on anaphora resolution accepted at COLING 1994 in Kyoto Japan, and again in 1994 while visiting a professor at KAIST in Taejon, South Korea, I wrote an internationally very well received paper on anaphora resolution for English to Korean Machine Translation. I continued working on anaphora resolution, and was invited to be keynote speaker at DAARC (the specialised event on anaphora resolution) three times. In 1996 and 1998 I published two influential papers introducing the so-called ‘knowledge-poor approach to anaphora resolution’ in which I proposed a new paradigm to this complex problem with the COLING’1998 (in Montreal, Canada) paper emerging as one of the highest ranking papers on the topic in terms of citations. Richard Evans was initially hired in 1998 to work with me on anaphora resolution: he and Constantin joined me in further exploring this challenging topic and developed several improved and optimised versions of my 1998 algorithm. Richard and Constantin are the authors of several important papers on this topic themselves. With Catalina Barbu, I developed the first evaluation workbench for evaluation of anaphora resolution algorithms which was a ground-breaking development. Other members of the group were also co-authors of papers on this topic, including but not limited to Le An Ha and Victor Pekar. My 2002 monograph ‘Anaphora resolution’ remains the most read and most influential reading on this topic.
Q: What do you predict for the future of the field of Natural Language Processing?
NLP programs will automate our life more and more. But my philosophy is for computer programs not to replace humans, only to assist them, and this is the philosophy I have been preaching in my career. I hope my philosophy will prevail.
Q: What do you think about the University’s recent “Transformation” proposals?
The vision of the University management of the 2000s for me was an example of institutional support for its best performers. The proposed closure of RGCL, SCRG and RIILP for me is an example of the opposite.
The fact that research has been targeted on a large scale (with many top researchers proposed to be made redundant) speaks of the short-sighted vision of our University management. University without strong research is doomed to have a bleak future. Consider the Research Institute of Information and Language Processing (RIILP) which is historically the most successful research institute of the University and for which I am Director. The University management has proposed to disband RIILP. Almost all staff of the Research Group of Computational Linguistics (RGCL) are proposed to be made redundant. This is a serious strategic mistake and can only harm the credibility of the university. RGCL has put the university on the map, many scholars know about the University of Wolverhampton because of RGCL, and many students come to Wolverhampton because of RGCL.
The fact that RGCL is currently in the red is listed as a reason to close it down. The University management seem to forget that RGCL has been the most successful department in terms of external funding: almost 25% of the whole University’s REF’2021 income comes from RGCL. Two successful large projects would fix the RGCL financial deficit. In closing down RGCL, the University would be losing one if its best assets and puts itself in an awkward position with regard to the international projects in which RGCL is involved.
Q: Is there anything else you would like to add to mark the 25-year anniversary of the Research Group?
The 25-year anniversary of RGCL reminds us how successful we have been during these 25 years establishing ourselves as internationally leading group, and how much we have done for the visibility of this university. It is sad to see nobody from the University managers congratulated us on this anniversary, and nobody congratulated us on any of the recent outstanding achievements which we reported in the daily news updates. It is sad to see what ‘recognition’ we are receiving from the University management.