Scott MacLean

PhD. Candidate, University of Waterloo

Email:	smaclean@cs.uwaterloo.ca
Office:	DC 2302D
Phone:	+1 519 888 4567 x35435
Mail Address:	Scott MacLean David R. Cheriton School of Computer Science University of Waterloo 200 University Ave. W. Waterloo, ON, Canada N2L 3G1

Welcome to my website! I am a doctoral student in the Symbolic Computation Group at the University of Waterloo. My supervisor is George Labahn.

MathBrush

I am a member of the MathBrush project, which researches approaches to building applications allowing intuitive input and manipulation of hand-drawn mathematics on PC tablets. I wrote the symbol recognition component of our prototype application, MathBrush, as a co-op student.

My Master's thesis work was a preliminary investigation into a new approach to semantic analysis, the phase of recognition that attaches semantic meaning to symbol arrangements. In particular, I investigated the view that mathematical handwriting is structured but uncertain, in the sense that we know it possesses a certain amount of structure, but it is not clear what all the small pieces are and how they fit together. I developed a preliminary parser that used fuzzy set theory to model recognition uncertainty and relational context-free grammars to model the formal structure of two-dimensional math syntax. The formal model captures all valid semantic interpretations of a users' writing, while the fuzzy sets allow for differing confidence for each interpretation.

Another aspect of my Master's work, which I wrote my thesis on, deals with issues related to recognition accuracy and sketch corpus creation. I analysed the requirements and practicalities of corpus creation, and offered a standard methodology to follow when creating ground-truthed corpora so that they may be maximally useful for training and evaluating recognition systems.

My doctoral research expands on the themes touched on in my Master's work. I am currently working on the following problems:

• applying Bayesian analysis to the multi-dimensional parsing problem
• improving the complexity of multi-dimensional parsing
• find other efficient and effective ways to intelligently resolve recognition ambiguities
• attempt to systematize the combination of formal descriptions and ambiguity models
• work to standardize measurement schemes for recogizer accuracy and establish recognition contests

Publications

Refereed publications

• S. MacLean, D. Tausky, G. Labahn, E. Lank, and M. Marzouk. Is the iPad useful for sketch input? A comparison with the Tablet PC. Sketch-based Interfaces and Modeling, 2011. Online appendix.
• S. MacLean, G. Labahn, E. Lank, M. Marzouk, and D. Tausky. Grammar-based techniques for creating ground-truthed sketch corpora. International Journal of Document Analysis and Recognition, Vol. 14, 2011. pp. 65-74.
• S. MacLean, D. Tausky, G. Labahn, E. Lank, and M. Marzouk. Tools for the efficient generation of hand-drawn corpora based on context-free grammars. Sketch-based Interfaces and Modeling, Seventh Symposium on, 2009. pp. 125-132.
• G. Labahn, E. Lank, S. MacLean, M. Marzouk, and D. Tausky. MathBrush: A system for doing math on pen-based devices. Document Analysis Systems, Eighth IAPR Workshop on, 2008. pp. 599-606.
• G. Labahn, E. Lank, M. Marzouk, A. Bunt, S. MacLean, and D. Tausky. MathBrush: A case study for pen-based interactive mathematics. Sketch-based Interfaces and Modeling, Sixth Workshop on, 2008. pp. 142-150.
• G. Labahn, S. MacLean, M. Marzouk, I. Rutherford, and D. Tausky. A Preliminary report on the MathBrush pen-math system. Maple 2006 Conference, 2006. pp 162-178.

Unrefereed publications

• S. MacLean, G. Labahn. Recognizing handwritten mathematics via fuzzy parsing. University of Waterloo Computer Science Technical Report CS-2010-13, 2010.
• S. MacLean, G. Labahn, E. Lank, M. Marzouk, D. Tausky. Toward math-literate computers. GRAND National Conference, 2010. (See also the associated poster.)
• S. MacLean and G. Labahn. Elastic matching in linear time and constant space. Document Analysis Systems, Ninth IAPR Workshop on, 2010 (short paper).
• S. MacLean. Techniques for creating ground-truthed sketch corpora. MMath thesis, University of Waterloo, 2009.
• S. MacLean. Matching techniques for mathematical symbol recognition. Internal report, Symbolic Computation Group, University of Waterloo, 2007.
• S. MacLean, G. Labahn. Segmenting handwritten mathematical expressions. Maple 2006 Conference, 2006.
• G. Labahn, S. MacLean, M. Marzouk, I. Rutherford, and D. Tausky. MathBrush: An experimental pen-based math system. Dagstuhl Seminar Proceedings.

Other talks

• S. MacLean. Tools for hand-drawn corpus generation. Pen-based Mathematical Computation, Third International Workshop on, 2009.
• S. MacLean, G. Labahn. Mathematical symbol recognition in the MathBrush system. CMS-MITACS Joint Conference, 2007.

Posters presented at local meetings

• S. MacLean. Comparing and combining ORCCA symbol recognizers. ORCCA Joint Lab Meeting (Dec. 2010).
• S. MacLean. Elastic matching in linear time and constant space. ORCCA Joint Lab Meeting (Jan. 2009).
• S. MacLean. Fuzzy parsing and fuzzy branchings. ORCCA Joint Lab Meeting (Dec. 2008).
• S. MacLean. A parsing algorithm for a class of relational context-free grammars. ORCCA Joint Lab Meeting (March 2008).
• S. MacLean, G. Labahn. Comparing Isolated Symbol Recognition Methods. ORCCA Joint Lab Meeting (2006).

Last modified in June, 2011