Professor Ruslan Mitkov is an expert in Computational Linguistics at the University of Wolverhampton. He is Director of the Research Institute for Information and Language Processing and Head of the Research Group in Computational Linguistics.
Research in this area recently scored highly in the Research Assessment Exercise (RAE 2008). The research of the group has been rated as internationally leading, internationally excellent and internationally recognised. According to league tables published in the Guardian, The Times and Research Fortnight, research in Linguistics at the University of Wolverhampton is one of the top six in the UK.
Computational Linguistics has to do with the processing of human languages by computers. This could involve understanding or translating. For example, imagine you have to find the answer to the question ‘When were potatoes first imported into Britain?’ If you do a keyword search, the search engine will return 100s of matches containing the words ‘potatoes’ and ‘Britain’ but will not give you the answer. But one of the technologies we developed, called Question Answering, understands your question and finds and provides the answer for you. It is worth pointing out that computers find it very difficult to understand human languages due to the ambiguity and irregularity of language – so if I said, “I saw John with binoculars” it would not understand if I had used the binoculars to see John, or I had seen John using the binoculars. Of course, ambiguity is a problem for humans too but it is much harder for computers.
One of the applications we have developed is called automatic summarisation. This is a program that can read a lot of pages in a second and generate a summary of the most important text for you. Another application is a program we are working on with the National Board of Medical Examiners (USA) where the system reads medical texts and generates multiple choice tests.
I enjoy the fact that Computational Linguistics, often referred to as Natural Language Processing when talking about applied research, could be used in many areas of life. Our priorities at the moment are to use our technology in healthcare and e-learning. An example of what we would like to develop is a tool that will assist people with dyslexia or dementia and make it easier for them to read complicated texts.
I am working on 20-plus topics at the moment! My favourite topics currently are the automatic generation of multiple choice tests and the use of Natural Language Processing in teaching foreign languages. In addition, we are developing tools that will improve the efficiency of translators. I am also well known for my work on anaphora resolution – this is the challenging task of the computer being able to identify references (eg. whom or what pronouns refer to) in text. I generate a lot of ideas and many of them I pass on to our PhD students and we work on them together.
I am engaging in more and more international collaborations, with a view to securing more external funding for research. I am looking forward to commercialising some of our products and the possibility of setting up a start-up company. I would not feel satisfied if my research was not beneficial to society, so my dream is to use our technology to improve medical care as a legacy to my late parents, who were both medical doctors and I owe them a lot in terms of my academic career.
Natural Language Processing cannot wave a wand and solve all the problems, but we would want to advance this area in the next five years. Little advances can make a real difference in this field, and the applied research we are involved in can be of practical use to many areas of society.
I think seeing people in my field cite my work and read my books. Another personal achievement is when I am invited to conferences and meet young researchers who have said they have wanted to meet me and that my work had helped them to develop their own research successes.
Hopefully the RAE results will motivate us even further, and help us to attract more funding for research and more top researchers. I am very pleased that the RAE success was not an individual but a collective success within the Research Group in Computational Linguistics and the Research Institute for Information and Language Processing.
I think the most unusual thing is that to be a success in this field you need to have a background in computer science, mathematics and linguistics. We collaborate with people from all kinds of backgrounds. The difficulty of computers understanding human language due to its ambiguity, irregularity and in general, complexity, is a major challenge.
Some people think that we are trying to replace humans but we are not – we are trying to develop programs which help humans. People also think that computers can do wonders, but they cannot. They are not as intelligent as humans and they should not be. Natural Language Processing should be used in specific scenarios and in specific domains. It is only then that it can be sucessful.