James Journal Presentation

From Semantic Portal Wiki

Presentation given at CSCI 6966 Advanced Semantic Web (Fall 2008)#Lesson 13

• Speaker: James Michaelis
• Title: Flink: Semantic Web technology for the extraction and analysis of social networks
• Authors: Peter Mika
• Conference: Journal of Web Semantics 3 (2), 2005
• URL: http://www.websemanticsjournal.org/papers/20050719/document7.pdf
• Date of Presentation: 2008/12/04

Questions

ID	Question	Name	Answer
Flink Ankesh	The ontology of research topics (shown in Fig 3) is very interesting. However I'm concerned with utility of its mention in this paper. Please correct me if I am wrong- but I feel that ontology generation is not part of the system. Also nothing about it is discussed in Sections 2 or later. So what do you think must have been the motivation of the author behind its inclusion?	Ankesh Khandelwal
Flink Jiao	In section 3.2, "Flink also makes use of the rule language for carrying out identity reasoning". Name matching and and object identification are discussed. Object identification baed on IFP really makes sense. About name matching, differences in the last names are disallowed. Does this mean differences in the first names are allowed? If this it true then "jiao tao" and "jia tao" would be viewed as same persons however they are not.	Jiao Tao
Flink Jiao2	In figure 4 of section 3.1, four knowledge resources, which are HTML pages from the web, FOAF profiles from the semantic web, public colections of emails and publications. It is said web mining is used to discover the social network extraction from given HTML pages. How about the other knowledge sources? I think web mining can also be used to discover the associaions between researchers, at least from bibliographic information. For example, coauthor relationship in a given publication is one kind of proof of the associations between the authors; and the keywords of the publication suggest the research topic/interestes of the authors. Did the authors mention these in the paper?	Jiao Tao
Flink Jiao3	In the "errors in the extraction of specific cases" of section 4, it is said that disambiguation terms are used to mitigate the possible ambiguity (i.e. two different persons have same name) in the knowledge collection stage.However, only two disambiguation terms, semantic web and ontology, are mentioned in this paper. Are they enough? At least in semantic web domain, there are some other words can be used, like description logic, semantic query, and semantic web service, etc.	Jiao Tao
Flink Jiao4	In section 4, the author mentiones several aspects which may affect the quality of the data they collected. "To verify our method, we also plan to execute a separate study, where we compare the results from a traditional questionnaire method to the acquisition methods described here." Did the authors have any continual work on this? And what is the result of this comparison study?	Jiao Tao
James Journal Presentation GTW 1	The so-called "Google Mindshare" metric discussed in section 3.1 uses co-occurrance page count of a set of names and a list of interests. This seems like a particularly dated way of computing such a metric, just three years on. What are some better ways you can think of to compute these values? How can alternative approaches make use of semantic data to ensure intentional co-occurance as opposed to co-occurances in which the names and interests just happen to be on the same page (with the system unable to determine the context of the co-occurance)?	Gregory Todd Williams
James Journal Presentation Joshua Shinavier 1	In the introduction, it is stated that user profiles in centralized social networking services "cannot be exported in machine processable formats". Why not? As is mentioned in the paper, there are several large producers of FOAF, for example, my.opera.com. It's not necessarily true that such systems "do not let users to control the information they provide on their own terms", either. In fact, centralized services may be some of the best potential sources for well-formed Semantic Web data.	Joshua Shinavier
James Journal Presentation Joshua Shinavier 2	The author shows how Flink might be used for social network analysis, then note that he plans to report on the results of his study of the SW community in a future publication. What are these results? Has Flink led to any future (post-2005) work?	Joshua Shinavier
James Journal Presentation Shangguan	In section 3.2, we read "We hope that in the future our storage facility will provide native features for context support, which would improve the efficiency of storing and querying such information...". What information should be considered a "context support", in the sense that adding this information will facilitate querying? Do we need some extra reasoning work on these additional context information?	Zhenning Shangguan