a journal of a researcher

Monday, June 28, 2004

Montonicity in Qualitative Modeling

Monotonicity analysis on scattered data is to determine the relation between two variables, i.e. increasing, decreasing or constant of one variable when another variable is increasing. It is an important feature to leverage when constructing qualitative model.

The existing approaches are rooted in classic data analysis, such as approximation, regression. It is a domain always attracting attentions. I’ve implemented Eanonn Keogh ’s methods and Martin Brooks’ method to compare their pros and cons in my application. A draft document summarizes my results on this problem.

Today, I asked a question on Keogh’s complexity. It is very delightful that I got answers from my NRC colleagues within minutes and they are right. The group of people I consulted are Martin Brooks , Peter Turney , Daniel Lemire and Joel Martin .

Thursday, June 24, 2004

A Researcher’s Visibility

Citation is always a criterion for evaluating a researcher. First, you have to do good work. Second, the problem you are dealing with is important enough to attract attentions. Third you have to publish in the main stream conferences or journals so that your papers are accessible. The later point is what I ignored.

Now, you have another method, the internet. Can your web site be googled by your name? and your paper are available on it? or even better, google ranks you the first by googling your domain. Associate the domain keywords with your website? Many tricks. But you have to wait before you are really worth it.

This is from my discussion with Daniel Lemire .

Blogging? Murmur?

More and more I am visible on the web, because my colleagues are indexing me and linking me. Sooner or later more people will visit my blog and my website. But I found I was trapped in someway.

First, is it really my journal? The more people visit my site, the more I have to be responsible for what I publish. When people want to publish before a group of audiences, people twist the truth. People still point me to my essay published on the Chinese largest newspaper the People’s Daily (overseas version) even several years later. But I am not proud of it, because I think my mind was twisted by simply appearing on the People’s Daily, no matter to say the revision from the editors (even with the fact that they are minor). My other Chinese writings now are well blocked in my computer or my real journal.

The first Chinese winner of Nobel Prize for Literature Gao Xingjian said, “writing is a kind of personal murmur”. Only the personal murmur is the true internal feeling of people. Do they expect to be heard? Sometime, otherwise, the writer would not publish his book. But I am sure he does not care how people think about his murmur, and he does not change the murmur because it will be heard.

To the people who are indexing my blog, are you too serious and too waste your time?

Tuesday, June 22, 2004

My homeage

Here is the personal website of Yuhong Yan .

Then I need RSS Feeds. Publish myself on the web? I am a shy people. I don't think I like people to check my thoughts. Well, though taking a sceptical attitude, I still give it a try.

Monday, June 21, 2004

Conflicts and Opportunities

I read again the paper "Characterizing Diagnoses and Systems" by Johan de Kleer, Alan Mackworth and Raymond Reiter. It is one of the four classic papers in this domain. The more I read it, the more I like it. When the paper was published in 1992, Canadian led the area. Reiter was from U of T and Machworth from UBC. But now I don't see Canadian in DX anymore. Reiter past away in 2002 and is memorized by AI people by devoting IJCAI03 proceedings to him.

I installed Tomcat on my notebook as a start to prepare my web services course. I didn't touch intensive programming for three years.

Today I also have a little misunderstanding with one of my collaborators. This makes me think what is important in my life and what is the leadership. On one hand, I take it seriously and enjoy all the joys this career can offer me; on the other hand, I do not take it as the only meaning of my existence and I won't do anything I can to get what I want. This misunderstanding orginally comes from my lack of communication before I take an action. Thinking always everyone's interests is a condition of leanership. Be generous and have vision are the other two conditions. That is what I learned today.

Thursday, June 17, 2004

New Publications

Two papers accepted

• Yuhong Yan, "Sensor Placement and Diagnosability Analysis at Design Stage", MONET Workshop on Model-Based Systems at ECAI 2004, August 22-26, Valencia, Spain.
• Yuhong Yan, Daniel Lemire and Martin Brooks, "Monotone Pieces Analysis for Qualitative Modeling", MONET Workshop on Model-Based Systems at ECAI 2004, August 22-26, Valencia, Spain.

Friday, June 11, 2004

I am a new adjunct

I am appointed as an adjunct professor at the faculty of CS in UNB. Adjunct is a synonym of a junk professor. How an adjunct not to be a junk?
First what the students need from an outsider?
Second what the professors need from an outsider?
Server this community and benefit from it.

Web services vs. Semantic Web Services

Web services become a special term for services using UDDI, SOAP, WSDL and BPEL. Web services have two advantages: another distributed system architecture and a service-based business model. Compared with other techniques for distributed system like CORBA, IDL, web services provides easy interoperation via Internet protocol. As a service model, web services can integrate different services into a business process. Currently Web services are more in information providing services, like stock price, flight information. A typical example is RSS feeds syndication. But its future is in e-business process, involving transactions.

Semantic Web Services focuses on using annotations to describe web services. Semantic Web Services has its specifications from W3C, i.e. RDF, OWL-S. The annotation is based on description logic and ontology. The assumption of semantic web services is that if we can clearly describe the services and their properties, we can use logic reasoning to enable functions such as transferring information between services, mapping ontologies. But it goes to the curse of AI: can people represent knowledge in an efficient way, formally and completely? The enumeration of properties the semantic web adopts can be only linear and domain dependent. This dooms the fate of semantic web. That is why few companies support OWL-S.

Thursday, June 10, 2004

Start blog

As a researcher, I am shame to publish my idea before it is mature. But I was attracted by my peers' blog. So I become a blogger and share my ideas within my circle.