• Home
• Projects
• Publications
• Misc

google scholar profile

Natural language dialogue

• "A Cross-cultural Corpus of Annotated Verbal and Nonverbal Behaviors in Receptionist Encounters," by Maxim Makatchev, Reid Simmons, and Majd Sakr. Workshop on Gaze in HRI: From Modeling to Communication, Boston, MA, USA, March 5, 2012. arXiv:1203.2299v1, the corpus is free.
• "Perception of Personality and Naturalness through Dialogues by Native Speakers of American English and Arabic," by Maxim Makatchev and Reid Simmons. SIGDIAL, Portland, OR, USA, June, 2011. An unabridged version: arXiv:1105.4582v1
  explain in plain English
  
  We can judge the personality of other people based on what they say. However, you and I may think differently about the same person. This is not surprising, considering all the possible differences between the two of us: native language, gender, and so on. In this study, we compared this kind of judgements made by people who natively speak American English and Arabic. We hired them over the internet and showed them texts of short English dialogues. We then asked them to score the personality of one of the dialogue speakers. They also rated how natural was what that speaker said. We found that Arabic and American English speaking participants are quite different when it comes to judging formal language. For example, American English speakers thought that formal greetings, answers to questions, and disagreements were not natural. Arabic speakers thought that those same dialogues were natural. They also thought that people who spoke formally during disagreements and apologies were open and conscientious, while American English speakers didn't think so. Finally, American English speakers thought that those who said "Sorry, I am not sure." were more agreeable than those who said "Sorry, I have no idea...." or "Sorry, I don't know." Arabic speakers didn't show this difference.
  explain like I'm five
  
  You're gonna creep Americans out if you say things like "Good morning, sir" and "I am afraid that is not correct." Saying the same things to people who natively speak Arabic, on the other hand, is OK. We used statistics, so it must be true.

Human-robot dialogue

• "Cross-cultural Believability of Robot Characters," by Maxim Makatchev. PhD thesis, Robotics Institute, Carnegie Mellon University, tech. report CMU-RI-TR-12-24, February, 2013. pdf
• "Expressing Ethnicity through Behaviors of a Robot Character," by Maxim Makatchev, Reid Simmons, Majd Sakr, and Micheline Ziadee. Proceedings of the 8th ACM/IEEE Int. Conf. on Human-Robot Interaction (HRI), Tokyo, Japan, March, 2013, pp. 357-364. A version with plots: arXiv:1303.3592
  explain in plain English
  
  This paper makes two main contributions. First, it expands the methodology for creating ethnic characters by incorporating a step that evaluates ethnic salience of behaviors via crowdsourcing. We show that the resultant behavioral cues of ethnicity do indeed affect ethnic attribution of robot characters across different appearances. Second, we use these ethnically salient characters to evaluate the hypothesis of ethnic homophily between humans and robots. Namely, the hypothesis that users who perceive the robots as ethnically similar to themselves will like the robots more and will perform better on a task. For now, we do not find support of this hypothesis.
  explain like I'm five
  
  Similarities attract. For example, people are attracted to virtual characters of similar ethnicity. However, we did not know whether people are attracted to robots of similar ethnicity. So, we created robots that express ethnicity by their behaviors and found that, in this particular case, it didn't make a difference on how attractive they are.
• "Believable Robot Characters," by Reid Simmons, Maxim Makatchev, Rachel Kirby, Min Kyung Lee, Imran Fanaswala, Brett Browning, Jodi Forlizzi, Majd Sakr. AI Magazine, AAAI, Vol.32, No. 4, Winter 2011, pp. 39-52. pdf
  explain in plain English
  
  Pioneer Disney animators had an idea of a believable character―someone who provides an illusion of life, a personage to whom the audience can relate. Similarly, believability is a goal in performing arts and literature: Stanislavsky used to say "I don't believe you!" Joseph Bates introduced this notion for on-screen agents. We argue that believability is important for human-robot interaction as well. We associate believability with social attitude, and show that users that treat robots socially are more persistent in the face of the robot's non-understandings, and as a result perform better on an information-seeking task. We discuss how to increase social response using storyline, verbal and non-verbal behaviors, and culturally-specific language in the context of 3 robot receptionist characters located in Pittsburgh, USA and Doha, Qatar.
  explain like I'm five
  
  People who treat robot receptionists in a more human-like way actually have better chance at getting what they want from the robot. This article is about how a robot character can facilitate this kind of behavior from humans.
• "Do You Really Want to Know? Display Questions in Human-Robot Dialogues," by Maxim Makatchev and Reid Simmons. AAAI Fall Symposium on Dialog with Robots, Arlington, VA, USA, November, 2010. abstract and pdf
  explain in plain English
  
  People ask questions for different reasons. Sometimes it is just to test the other person's knowledge. Linguists call this kind of questions display questions, since they are aimed at displaying someone else's knowledge. Similar thing happens when people talk to a robot. How to measure this? One way would be to make the robot ask the user something like "Are you really testing me?" Of course, the user may not answer truthfully. So, instead, the robot asks "Do you know the answer? Can you tell it to me?" This way we found that at least 16.7% of users ask our robot receptionist display questions. Now that we have some examples, we can try training the robot to recognize such questions and respond accordingly.
  explain like I'm five
  
  If you leave a robot on its own and let everyone talk to it, you're gonna see some weird stuff. Like people typing their own names. Future robots should recognize that and make snarky remarks. Yes, it's a must. But first we should get some examples of such behavior. We figured out one way of doing that and wrote a paper.
• "Dialogue Patterns of an Arabic Robot Receptionist," by Maxim Makatchev, Imran Fanaswala, Ameer Abdulsalam, Brett Browning, Wael Ghazzawi, Majd Sakr and Reid Simmons. Late-breaking report at Int. Conf. on Human-Robot Interaction (HRI), Osaka, Japan, March, 2010. abstract and pdf
• "Incorporating a User Model to Improve Detection of Unhelpful Robot Answers," by Maxim Makatchev and Reid Simmons. Int. Symp. on Robot and Human Interactive Communication, RO-MAN'2009, Toyama, Japan, September, 2009, pp. 973-978. abstract and pdf, extended version, slides
• "How do people talk with a robot? An analysis of human-robot dialogues in the real world," by Min Kyung Lee and Maxim Makatchev. Proc. of Int. Conf. on Human factors in computing systems, CHI'2009, April, 2009. pdf
• "Relating initial turns of human-robot dialogues to discourse," by Maxim Makatchev, Min Kyung Lee, and Reid Simmons. Late-breaking report at Conf. on Human-Robot Interaction, HRI'2009, March, 2009. pdf

Anomaly detection

• "Learning Outbreak Regions in Bayesian Spatial Scan Statistics," by Maxim Makatchev and Daneil B. Neill, Proc. ICML Workshop on Machine Learning for Health Care Applications, Helsinki, Finland, July 2008. pdf

NL understanding for tutorial dialogue systems

• "Combining Bayesian Networks and Formal Reasoning for Semantic Classification of Student Utterances," by Maxim Makatchev and Kurt VanLehn, Proc. Int. Conf. on AI in Education, AIED2007, Los Angeles, July 2007. pdf
• "Understanding Complex Natural Language Explanations in Tutorial Applications," by Pamela Jordan, Maxim Makatchev, Umarani Pappuswamy. Proc. Workshop on Scalable Natural Language Understanding (ScaNaLU), HLT/NAACL, New York City, June 2006, pp. 17-24. pdf
• "A Natural Language Tutorial Dialogue System for Physics," by Pamela Jordan, Maxim Makatchev, Umarani Pappuswamy, Kurt VanLehn and Patricia Albacete. Proc. Int. FLAIRS Conf., Florida, 2006. pdf
• "Representation and Reasoning for Deeper Natural Language Understanding in a Physics Tutoring System," by Maxim Makatchev, Kurt VanLehn, Pamela W. Jordan and Umarani Pappuswamy. Proc. Int. FLAIRS Conf., Florida, 2006. pdf
• "Analyzing Completeness and Correctness of Utterances Using an ATMS," by Maxim Makatchev and Kurt VanLehn. Proc. Int. Conf. on AI in Education, AIED2005, Amsterdam, Netherlands, IOS Press, pp. 403-410. pdf
• "Mixed Language Processing in the Why2-Atlas Tutoring System," by Maxim Makatchev, Brian S. Hall, Pamela W. Jordan, Umarani Pappuswamy, and Kurt VanLehn. Proc. of a Workshop on Mixed Language Explanations in Learning Environments, AIED2005, Amsterdam, Netherlands. pdf
• "Relating Student Text to Ideal Proofs: Issues of Efficiency of Expression," by Pamela W. Jordan, Maxim Makatchev, Umarani Pappuswamy. Proc. of a Workshop on Mixed Language Explanations in Learning Environments, AIED2005, Amsterdam, Netherlands. pdf
• "Abductive Theorem Proving for Analyzing Student Explanations to Guide Feedback in Intelligent Tutoring Systems," by M. Makatchev, P. W. Jordan, K. VanLehn. Journal of Automated Reasoning, Special Issue: Automated Reasoning and Theorem Proving in Education, vol. 32, issue 3, pp. 187-226, 2004. A corrected draft is available: pdf.


• "Modeling Students' Reasoning about Qualitative Physics: Heuristics for Abductive Proof Search," by Maxim Makatchev, Pamela W. Jordan, and Kurt VanLehn. Proc. Int. Conf. on Intelligent Tutoring Systems, Maceió, Alagoas, Brazil, August 30-September 3, 2004, Springer LNCS, vol. 3220, pp. 699-709. pdf
• "Combining Competing Language Understanding Approaches in an Intelligent Tutoring System," by Pamela W. Jordan, Maxim Makatchev, and Kurt VanLehn. Proc. Int. Conf. on Intelligent Tutoring Systems, Maceió, Alagoas, Brazil, August 30-September 3, 2004, Springer LNCS, vol. 3220, pp. 346-357. pdf
• "Abductive Proofs as Models of Students' Reasoning about Qualitative Physics," by Maxim Makatchev, Pamela W. Jordan, Umarani Pappuswamy and Kurt VanLehn. Proc. Int. Conf. on Cognitive Modelling, Pittsburgh, July 30-August 1, 2004, Lawrence Erlbaum Associates Publishers, Mahwah, New Jersey, pp. 166-171. pdf
• "Abductive Proofs as Models of Qualitative Reasoning," by Maxim Makatchev, Pamela W. Jordan, Umarani Pappuswami and Kurt VanLehn. Int. Workshop on Qualitative Reasoning, Evanston, Illinois, USA, August 2-4, 2004, pp. 11-18. pdf
• "Abductive Theorem Proving for Analyzing Student Explanations," by P. W. Jordan, M. Makatchev, K. VanLehn. Artificial Intelligence in Education. Proc. of Int. Conf. on AI in Education, AIED 2003, Sydney, Australia, IOS Press, pp. 73-80. pdf
• "Extended Explanations as Student Models for Guiding Tutorial Dialogue," by P. W. Jordan, M. Makatchev, U. Pappuswamy. Natural Language Generation in Spoken and Written Dialogue, 2003 AAAI Spring Symposium Technical Report, SS-03-06, pp. 65-70. pdf
• "The Architecture of Why2-Atlas: A Coach for Qualitative Physics Essay Writing," by Kurt VanLehn, Pamela W. Jordan, Carolyn P. Rose, Dumisizwe Bhembe, Michael Boettner, Andy Gaydos, Maxim Makatchev, Umarani Pappuswamy, Michael Ringenberg, Antonio Roque, Stephanie Siler, and Ramesh Srivastava. Proc. 6th Int. Conf. Intelligent Tutoring Systems, Biarritz, France and San Sebastian, Spain, June 2-7, 2002, Springer LNCS, vol. 2363, pp. 158-167. pdf

Modeling, control and human-robot interface for mobile robots

• "System Design, Modelling, and Control of a Four-Wheel-Steering Mobile Robot," by Maxim Makatchev, John J. McPhee, S. K. Tso, Sherman Y. T. Lang. Proc. 19th Chinese Control Conference, Hong Kong, China, December 6-8, 2000, pp. 759-763 . ps.gz (843K)
• "Human-Robot Interface Using Agents Communicating in an XML-Based Markup Language," by Maxim Makatchev, S. K. Tso. Proc. IEEE Int. Workshop on Robot and Human Interactive Communication, RO-MAN 2000, Osaka, Japan, September 27-29, 2000, pp. 270-275. pdf (557K)
• "Cross-Coupling Control for Slippage Minimization of a Four-Wheel-Steering Mobile Robot," by Maxim Makatchev, Sherman Y. T. Lang, S. K. Tso, John J. McPhee. Proc. International Symposium on Robotics, Montreal, Canada, May 14-17, 2000, pp. 42-47. ps.gz (756K)

Complexity of picture language recognition

• "On the Complexity of Image Processing and Pattern Recognition Algorithms," by Maxim Makatchev and Sherman Y. T. Lang. Proc. Int. Workshop on Image, Speech, Signal Processing and Robotics, Hong Kong, September 3-4, 1998, vol. 1, pp. 217-222. ps.gz (117K). Also presented at the International Conference on Theoretical Computer Science, in Honour of Prof. Manuel Blum's 60th Birthday, 20-24 April, 1998, City University of Hong Kong.

Talks

• "On the Complexity of Image Processing and Pattern Recognition Algorithms," by Maxim Makatchev and Sherman Y. T. Lang. International Conference on Theoretical Computer Science, in Honour of Prof. Manuel Blum's 60th Birthday, 20-24 April, 1998, City University of Hong Kong.

Miscellaneous

• "BlindAid: An Electronic Travel Aid for the Blind," by Sandra Mau, Nik Melchior, Maxim Makatchev and Aaron Steinfeld. Tech. report CMU-RI-TR-07-39, Robotics Institute, Carnegie Mellon University, May, 2008. pdf
• "Modelling and Control of a Four-Wheel-Steering Mobile Robot," by Maxim Makatchev, John J. McPhee, S. K. Tso, Sherman Y. T. Lang. (Draft). ps.gz (449K)
• "On the Cell-based Complexity of Recognition of Bounded Configurations by Finite Dynamic Cellular Automata," by Maxim Makatchev. (Draft). arXiv:cs/0210009v1.