IWIS Seminar Autumn 2008

Peter Dolog's picture
04/25/2009 - 07:51
Etc/GMT

On August 22, 2008, IWIS group starts a series of research seminars intended especially for PhD and Master students. The seminar will take place ones in every 4 weeks. The seminar topic is World Wide Web. Each PhD student, master student and reasearch assistant (participants) selects a sub topic within the main seminar theme. The topic will be discussed on the seminar. The participants get assigned papers within the topic of interest. The rule of thumb is that the topic is slightly different from everyday project work. They will read papers and prepare a seminar presentation. The main goal is reading the articles in the broader scope of IWIS group research topics, improve the presentation and communication skills, expand on own knowledge and competence beyond the area studied in the PhD/Master project, as well as socialize. Deep discussions on the presented topics will be facilitated at the seminar as well.

Organization and Tutoring: Peter Dolog

Seminar Schedule, room 5.2.32
date day time program
August 22, 2008 Friday 13:00 - 15:00 Introduction by Peter Dolog, KIWI by Karsten Jahn, Personalization by Fred Durao, idSpace by Yujian Lin
September 19, 2008 Friday 13:00 - 14:00 Machine Learning by Yujian Lin, topic and papers assignement
September 24, 2008 Wednesday 13:00 - 14:00 Special Guest: Prof. Klaus Schmid, Flexible production of software systems - a product line approach
October 17, 2008 Friday 13:00 - 14:00 Reasoning and Reason Maintanance by Fred Durao based on paper from KIWI project: State of the Art: Reasong Maintanance: http://wiki.kiwi-project.eu/multimedia/kiwi-pub:KiWi_d2.3_final.pdf, Scientific paper writing by Peter Dolog
December 8, 2008 Monday 10:00 - 11:00 Reasoning and Ontologies for the Semantic Web by Karsten Jahn based on paper from Rasoning Web Summer school writtent by Thomas Eiter, Giovambattista Ianni, Thomas Krennwallner, and Axel Polleres: http://dx.doi.org/10.1007/978-3-540-85658-0_1

 

AttachmentSize
Yujian_Introduction_to_Machine_Learning.pdf93.07 KB
reasoning_maintenance.pdf198.31 KB
Peter Dolog's picture

seminar on 17 October 2008

We had various reflections on the trips and also last week where the IWIS group hosted two research visitors in idSpace project. It was interesting to see especially the reflections on this week from Yujian and Rory and how different creativity sessions helped them to get more understanding as well as envision further work in the idSpace project. The seminar featured a talk of Fred Duaro where he presented an introduction to the reasoning and reason maintanance systems with an exercise for us at the end. The talk illustrated principles and various types of reason maintanace systems very nicely so it served its purpose. The exercise Fred prepared was actually very good, as it showed also not just to test whther the audience (us ) understood what he was talking about but also to discuss othe issues such as negation as failure, fixing the inconsistency vs. just signalling it and so on. I presented shortly how I see a good scientific paper which should be based on a concrete problem, idea or proposal how to solve it, realization of the idea, evaluation of the realized idea as well as comparison to related work. This need was triggered first by Yujian but lately also by Karsten. We had a guest, Rory Sie, from the open university of the netherlands who also joined us. I think the seminar was again a very good educational activity for all of us.

Fred Durao's picture

Reason Maintenance Summary

In my presentation I showed the main concepts about Reason Maintenance. Basically Reason Maintenance provides basis for both explanations and revision/versioning for Reasoning (the step before maintenance). Inconsistency detection, calls for explanations, revisions/versioning and fault diagnosis are the typical benefits acquired from Reason Maintenance.

yjlin's picture

deductive reasoning V.S. inductive reasoing

I think it is a little bit difficult to tell the difference between deductive reasoning and inductive reasoning.
Deductive reasoning:
Premise1: All men are mortal;
Premise2: Socrates is a man'
Conclusion: Socrates is mortal.

Inductive reasoning:
Basic facts: Sun rises in the east and sets in the west everyday until today.
Conclusion: Sun will rise in the east and set in the west tomorrow.

At first glance, it seems not much difference among these two reasoning. The deductive reasoning is that first you have a true statement, then you have a assumption, then you reach the conclusion. It can be stated like this: We know All men are mortal, "if" Socrates is a man, "then" Socrates is mortal. Deductive reasoning have THREE statements in all, and has another name "syllogism".
However, Inductive reasoning is more like that the conclusion comes from a large number of particular examples (facts).

Anyway, it is very interesting to do some research on Logic and intelligence.

Peter Dolog's picture

seminar on September 19 2008

We have heart various possibilities for machine learning especially in connection to Web from Yujian. An interesting project of clustering of chineese web pages was presented and problems with chinese sybmbols in machine processign discussed. Fred presented shortly on KIWI tag based recommendation and at least two common topics with Yujian have been identified which could lead to some publications. I reported on my trip to SPLC 2008 and my guest talk at the SOAPL 2008 workshop. We spent 2hours on discussions instead of originaly planned 1 hour  but then again, discussions are always good if they lead to an interesting research. A very good seminar I must say from my side as an advisor 

yjlin's picture

seminar on September 19, 2008

At this seminar, I was impressed by two things:
1. the compensation in Web Services by Peter. Actually, this was the first time I heard about compensation in Web services, although I know of transactions in database. What's more interesting was that during the evening, a Chinese guy from the database group inquired me something about his danske bank account because he found that the available money was inconsistent with his balance. I said to him that if he has consumed with his card, that was possible the available amount would be less than the balance. Then he told me, he tried to buy a plane ticket, but he failed, how could the plane-ticket services subtract the money from his account. Then I comforted him and explained to him, trying from the web service perspective. and I told him there should be a compensation later, specifically a rollback operation in database under this situation. At the end, he appreciated the IWIS was really practical and then I played a joke with him that he might join our group if Peter had vacancies. Anyway, that was really a nice experience :-)
By the way, Peter, that would be great you presented the topic again at our seminar if convenient and time permits.

2. the relationship "similarity between different users" from Fred's presentation.
That is a concrete example of social web computing, or, web reasoning. I think that is a good idea to call in some services provided by dictionary service provider.
Second, the basic idea of this recommend mechanism Fred referred to is if User A is some kind of "similar" with User B then B might be also interested in what A is interested in. and four tag-based approaches are presented to define the similarity.
well, then some mechanism to define "interested in" should also be designed, I think.