MAIN | RESEARCH | PUBLICATIONS | SOFTWARE | APPLETS | BLOG
B.Sc. (Toronto), M.Sc. (Toronto), Ph. D. (Polytechnique)
Also available as RSS.
You might want to visit my blog.
October 16th 2005
This month, Steven Keith will be presenting our paper Analyzing Large Collections of Electronic Text Using OLAP during APICS 2005 in Wolfville (Canada).
September 6th 2005
A paper at IEEE Data Mining (ICDM-05) called An Optimal Linear Time Algorithm for Quasi-Monotonic Segmentation was accepted with Martin Brooks et Yuhong Yan. Only 22% of the papers were accepted. Last August, one of my articles was published by the International Journal of Interactive Technology and Smart Education: Collaborative Filtering and Inference Rules for Context-Aware Learning Object Recommendation with Harold Boley, Sean McGrath et Marcel Ball.
April 29th 2005
AAAI05 accepted our paper Quasi-monotonic segmentation of state variable behavior for reactive control (with Will Fitzgerald and Martin Brooks). The acceptance rate was 27%.
March 16th 2005
IJCAI05 accepted our paper Scale-Based Monotonicity Analysis in Qualitative Modelling with Flat Segments (with Yuhong Yan and Martin Brooks). The acceptance rate was 18%.
January 14th 2005
I gave a talk at the "Ateliers du CIRTA" (an eLearning institute) on rating-based collaborative filtering and presented briefly Slope One Collaborative Filtering.
December 13th 2004
With Anna Maclachlan (now at Idilia Inc.), we got our paper "Slope One Predictors for Online Rating-Based Collaborative Filtering" accepted at SIAM Data Mining 2005. The work reported in this paper served as the foundation for the inDiscover Web site. It describes a very simple, yet reasonably accurate approach to collaborative filtering which is especially easy to implement. The PDF version of the paper is available.
November 19th 2004
After founding SWIG, we organized the first big SWIG event: SWIG'04.
November 4th 2004
Our Web site inDiscover is now a Bell Sympatico/MSN site. Bell Canada is one of the largest Canadian company. The web site is a direct result of my research.
October 6th 2004
I gave a talk at the RALI lab. (University of Montreal) on Rating-Based Collaborative Filtering. Thanks to Guy Lapalme for the invitation. I talked about my work with Anna Maclachlan which lead to inDiscover.
September 21st
I'm giving a talk at the Language Industry Association (AILIA) in Ottawa: Semantic Web in Brief. I talk about what the Semantic Web really is, that is RDF. I give a brief overview of RDF with applications, but with little to no mention of XML.
September 8th 2004
With Yuhong Yan and Martin Brooks, we got another conference paper accepted (MBR'04): Monotonicity Analysis for Constructing Qualitative Models.
June 12th 2004
With Yuhong Yan and Martin Brooks, we published a new paper (MONET Workshop): Monotone Pieces Analysis for Qualitative Modeling.
May 29th 2004
I was a reviewer for Sreeraman Rajan's Ph.D. thesis supervised by professors Doraiswami and Stevenson, in Electrical and Computer Engineering at the University of New Brunswick. The thesis was on phonocardiograms classification using Morlet wavelets and is entitled A Wavelet-based Correlator Approach for Segmentation and Classification of the First Two Heart Sounds in a PCG Signal. In my humble opinion, it is a very good thesis properly defended.
May 13th 2004
One of the world's expert in e-Learning, Stephen Downes, took notice of our project indiscover.net in his blog.
OLDaily May 13, 2004 by Stephen Downes ----------------------------------------------------------- InDiscover Those of you who heard my talk in Winnipeg last week (or viewed the slides online) will recall my discusion of recommender systems. InDiscover is one such site, a service designed to help bring independent music to new listeners by recommending songs based on users' tastes. The site is run using racofi technology, developed by NRC in Fredericton in conjunction with the Sifter Filter project I described in my talk, and so it gives you a pretty good idea of how such a system would work for learning resources. What I also like about this site is its focus on new and independent musicians, showing how such systems will open the marketplace. Still looking for the RSS metadata feeds from the site, but hey, one thing at a time. The site could also use more songs, so if you know of independent artists looking for exposure, forward them this item. By Various Authors, May, 2004 http://indiscover.net/
April 24th 2004
I'm not longer a researcher at the National Research Council of Canada in Fredericton. I've left for a professorship at the University of Québec. It is with regret that I leave the Internet Logic Group, but I remain forever an NRC person. One big difference with my new life in Montréal is that they serve cappucinos on the first floor of the building I work in, as well as at every corner in the neighbourhood. I'm on the corner of Saint-Denis and Mont-Royal which is one of the best place to be in Montréal.
If you are looking for a good Ph.D. program, you might want to check out UQAM's informatique cognitive program (French). There is something to be said about a doing your Ph.D. in one of the most exciting cities in the world.
April 17th 2004
Our Music recommender site, inDiscover is on-line. I think that Sean did a fabulous job on it. The part where I know we got it right is that we ask you to rate independent music and thus, you can listen to the MP3s on-line for free.
April 1st 2004
Our e-Learning Summit talk last Tuesday on the potential benefits of rating learning objects was well received and generated lots of interesting debate. A few people insist that objects should be described objectively using a centralized top-down approach, but I think, I hope, it is a minority. I'm currently working on the corresponding paper and I have very little time left!
In related news, Sean McGrath received a NSERC USRA grant and he will be continuing work on the inDiscover collaborative filtering site. This work is part of the RACOFI Composer project lead by myself and Harold Boley.
In somewhat unrelated news, I'm in the process of remaking the Lemur OLAP library using a somewhat simpler API. Currently, I can already support data cubes well into the 100 GB range. While the code is in C++, I have a Python wrapper already and Owen Kaser seems interested in the possibility of an OCAML wrapper. The new version is still not publicly available, though we will release it some time this summer, I expect.
March 17th 2004
As an invited speaker, I gave an introduction to wavelets for Dan Kucerovsky's students.
March 16th 2004
Our proposed communication for the Learning Object Summit was accepted. The meeting will be on March 29th and 30th in Fredericton (New Brunswick) and is funded by Canarie and the project eduSource.
The title of our talk is Collaborative Filtering and Inference Rules for Context-Aware Learning Object Recommendation.
March 15th 2004
Nathalie and Jae updated Kian's home page.
March 8th 2004
The last two months were tough. I submitted 3 papers. First, there is a new paper with Anna Maclachlan, my favorite linguist on a simple but efficient technique for collaborative filtering: Slope One. Essentially, it is sufficient to compute the average difference between the ratings any two items got, and then you can use our really fast algorithms. I always work with Sean McGrath on the RACOFI Composer projet and he will test these brand new techniques on a live web site.
Also, I worked with Owen Kaser on a local moments buffering technique for very large data sets. The technique is called Ola. The paper also present a technique applicable to all algebraic queries called Hierarchical Bin Buffering. All of this can be very useful in statistics, visualization, and for very large databases (OLAP). We also finished an extended version of our paper Attribute Value Reordering for Efficient Hybrid OLAP already presented at DOLAP'03.
Finally, partly because of Sean and Stephen Downes, I got into RSS feeds. Shortly, my own site will have RSS feeds. It is already the case for the Wavelet Forum. I also wrote a module for jnews allowing me to use regular expressions to filter RSS feeds.
January 21st 2004
Tim Scammel broke the Planetary Society's Mars Lander puzzle today. He gave me some credit for having found the basis for the number system they used: 3. I found the number by seeking the smallest common factor.December 4th 2003
I was a guest of IEEE and the Computer Science and Statistics Department at UNB/SJ. I talked about wavelets, giving a historical overview and showing that you could do wavelets without using advanced mathematics.
November 22nd
I am the proud father of a healthy boy!
October 20th 2003
I gave a talk to representatives from UPEI, Université de Moncton, and UNB as part of the Arts Netlantic network. The talk was about the Racofi Music web site and the use of subjectivity in semantic web.
October 16th 2003
I'm back from Halifax. The stormy weather over there did some serious damage. I checked out for the first time Nathalie Saint-Jacques' new home. I went by Wolfville to give an invited talk where I met some of my old colleagues as well as some people I didn't know. I was also an invited speaker in André Trudel's Semantic Web course. In Halifax, I participated to COLA'03 which was a success. I also gave an invited talk at Acadia on collaborative filtering.
I've been quite busy lately. The RACOFI project was a success in that we've developed an excellent partnership with KnowledgePool and I've merged the output of my research with the RuleML/JDrew people: Bruce Spencer and the famous Harold Boley. (Isn't that yucky? Do research with a company... beurk... can't have that! -- ironic statement) This will allow us to have our collaborative filtering technology embedded quickly in software to be used by the e-Learning industry. We've also had a paper accepted at IEEE COLA'03 regarding this work. I will present it in October in Halifax. Also, my paper on scale and translation invariance collaborative filtering has been accepted for publication in the Journal of Information Retrieval. All and all, this is a surprisingly successful venture. I'm now trying to put more solid mathematics behind it all and hopefully, I'll make some progress in the coming weeks.
I will also try to branch in the Howler project and my work with Owen Kaser. Mining ratings using OLAP appears to be a worthwhile idea. Talking about Owen brings me to the bit of news about Owen and I getting our Attribute-Value Reordering paper accepted at DOLAP 2003. Unfortunately, I won't attend DOLAP 2003, but Owen will be there. I'm currently trying to convince Owen to work with me on scalable range queries... Not sure I will succeed, but I will try!
I've begun working on a project related to contour trees with Martin Brooks. I'm trying to convince Martin to use CVS for collabortion like I did with Harold and Owen. Hopefully, this will work. CVS is just such a great tool.
In personal news, our cat Yahoo has disappeared for I don't know how long. Milou claims the cat is just lost. I fear that he may have gotten stuck somewhere is died. We did a search in the area, but because we live in the woods, there is only so much we can do. I'm afraid I might never see my cat ever again. Yahoo was a great friend.
Jolie, our 2 years old cat, had disappeared for the last 4 days. When Milou and I finally decided to go take a walk nearby yesterday, we heard her cries for help. We quickly found her in a big tall tree. She was stuck high in the air and wouldn't come down. All attempts to get close to her using a ladder only pushed her to go upward. We tried to get help from the SPCA, the vet, the firemen,... nobody would help us. Milou eventually stumbled on a tree climber called Mr Tree using the Yellow Pages. The guy came quickly and retrieved Jolie without even using a ladder. It was very impressive. It costed me $200, but Jolie is safe at last! (Happy faces.)
I was at the Summer meeting of the CMS in Edmonton together with Martin Brooks, Serge Dubuc, Rémi Vaillancourt and others. We held an Approximation Theory workshop organized by Bin Han and Rong-Qing Jia. I also met some Acadia University folks including Nancy Clarke and Franklin Mendivil.
The web site for the Grad course on OLAP I teach with Owen Kaser is up.
I'm attending the 71st annual ACFAS meeting in Rimouski, Quebec. I'll be talking about high resolution local interpolation. The ACFAS stands for Association francophone pour la savoir or the "French-speaking Society for Knowledge" (rough translation). Rimouski is a beautiful city and from my hotel room, I've got a great view of the Saint Lawrence river.
This summer, I'll be teaching a course on Advanced Technologies for E-Business at the University of New Brunswick. I will be discussing OLAP and Owen Kaser has agreed to teach with me.
I am in Moncton at the 69ième colloque des sciences mathématiques du Québec (69th Quebec Mathematical Sciences Colloquium). I came to talk about multistep subdivision schemes. It is a very good meeting.
Finally, yesterday I bought some 802.11b gear and Milou is surfing in the living room. It took me about 5 minutes to set it up and no swearing. Wireless networking is finally ready for everybody.
Two videoconference meetings with Martin Brooks this week. I'm really into his quasi-monotone theory. It is very nice, but I'm looking forward to see how it works in applications.
My work on collaborative filtering with student Nancy Howse was integrated in the Sifter Project with Mosaic Technologies Inc. This is very exciting as Nancy will get to spend the summer with us working on the filtering of e-Learning objects. It will also get me a chance to challenge my views on collaborative filtering.
This has been a very productive month. I got some NSERC funding. I taught a few classes on multiscale methods for the solution of nonlinear ODEs at UNB for John Stockie. I'm also making very good progress in many of my research projects and, in particular, on OLAP. There is also on piece of good news I'm very excited about by can't quite share with the world yet.
It was my birthday a few days ago. My wife offered to me the Lord of the Ring novels. Isn't she sweet!
My paper on Multistep Subdivisions Schemes was accepted for publication in the Curves and Surfaces 2002 Proceedings. I think that it is an important paper.
It looks like we are going to have another war in Iraq. Last time it happened, I was in Toronto wearing a black cloth around my left arm as a sign of protest. I may well do it again. Apparently, Iraq would be responsible for 9/11 or some other recent terrorist events. I don't see any proof of it. I believe Iraq is run by violent and crazy men, I don't believe it is a land of evil. I believe it is a land of oil. "How Bush and his junta succeeded in deflecting America's anger from bin Laden to Saddam Hussein is one of the great public relations conjuring tricks of history." (John le Carré, Times/UK, January 15, 2003 in The United States of America has gone mad).
I joined the Fredericton-based Internet Logic group lead by Bruce Spencer and with Harold Boley, Sandy Liu, and Scott Buffet as collaborators. My research will be on Data Mining for e-Commerce and it complements the work done by the other researchers on Semantic Web, Web Services, and Utility Theory. Verlé Harrop is replacing me as Group Leader for e-Health in Saint John. Good luck Verlé!
I received an unexpected and extraordinary gift from Franco Lenardon (CERN): a pile of books! Thanks Franco!
I gave a talk at the University of New Brunswick (Saint John), it was a shorter version of my recent talk in Ottawa.
I gave a talk at the NRC-IIT Colloquium in Ottawa: Life behind a web portal: Large Data Sets, Approximation Theory, and Wavelets.
I was at the 5th International Conference on e-Commerce Research in Montreal. The conference was very interesting and I met great people.
I gave a talk at the Math. & Stat. Department at the University of New Brunswick with the title: High Resolution Subdivision Schemes.
The University of New Brunswick granted me Adjunct Professor status. This recognition is at the Associate Professor level.
I am at CASCON 2002. IBM gave me a brand new laptop as for my Best paper award.
I am now the Team Leader for the eHealth Research Group at the National Research Council of Canada.
My paper Wavelet-Based Relative Prefix Sum Methods for Range Sum Queries in Data Cubes has won the Best Paper Award!
Gilles Deslauriers, my Ph.D. thesis supervisor dropped by Fredericton yesterday. After a great career as a mathematician, he is now retired and lives on a large piece of land somewhere in Quebec.
My paper on wavelet-based OLAP has been accepted by CASCON 2002. CASCON is a major Computer Science conference held in Toronto and organized by IBM Research and the NRC.
I'll be at the "Curves and Surfaces" conference in Saint-Malo until early in July. We'll also visit the Rennes area.
We've moved near UNB where the NRC E-Business research group is located. I've visited the new center and it looks really exciting! The Canadian government is really serious about this new research group and all the necessary ressources are invested to make this center a key player in world-wide E-Business research.
Classes have ended. Most people wrote their finals by now. I was particularly happy with the results I got in Numerical Methods and I think students enjoyed the course. This was a fun year!
Unfortunately, I won't be back next year at Acadia. While this is certainly one of the best mathematics department in Canada, I've been offered a position as a researcher at that National Research Council of Canada (NRC) where I'll be doing more leading-edge research in data processing and telehealth. At least one of my students got a job at the new e-business research group where I will be working, so there will be at least one familiar face!
My abstract on "High Resolution Subdivision Schemes" has been accepted for the June-July "Curves and Surfaces" meeting in Saint-Malo. This meeting is one of the most important meeting in Approximation Theory this year.
Debbie Boutillier from the Mathematics Department at Acadia just won this year student's prize for excellence in teaching! My congratulations Debbie!
I just marked the first assignment for the Numerical Methods 2 class. I'm quite impressed by the skill of the students. Next, we will work on a project proposed by Richard Karsten. We even have a special class this Friday where Richard will motivate the problem! Meanwhile, we continue our work on interpolation with Hermite, Newton, Lagrange, Deslauriers, Dubuc, Dyn, Levin, Gregory...
Life in Grand-Pré. One of the nicest little place in Canada. There is but one problem with it. The snow! I had to spend all day stuck inside with 10 cm of snow on the only road out of my house. Finally, at exactly 3:30 pm, the plow came by to open Grand-Pré road. See picture.
In december, the Quebec Mathematical Society (Association Mathématique du Québec) has published an interview with me concerning my work in industrial mathematics.
Publication of an article on wavelets (in French) in JDNet giving the following URL as a reference to the readers www.ondelette.com/lemire/onde.html (my page on wavelets!).
MAIN | RESEARCH | PUBLICATIONS XML| SOFTWARE | APPLETS | BLOG XML
© 2001-2006, Daniel Lemire (lemire at acm dot org)
