Debbie Rank Typed Graph Walks Presentation

From Semantic Portal Wiki

Presentation given at CSCI 6966 Advanced Semantic Web (Fall 2008) - Lesson 9

• Speaker: Debbie Heisler
• Title: Learning to Rank Typed Graph Walks: Local and Global Approaches
• Authors: Einat Minkov and William W. Cohen
• Conference: 9th WebKDD and 1st SNA-KDD 2007 Workshop on Web Mining and Social Network Analysis
• URL: http://portal.acm.org/ft_gateway.cfm?id=1348550&type=pdf&coll=GUIDE&dl=GUIDE&CFID=6937796&CFTOKEN=74138124
• Date of Presentation: 2008/10/23

Questions

ID	Question	Name	Answer
Debbie Rank Typed Graph Walks Presentation Gregory Todd Williams 1	The formalization of the lazy graph walks in section 3.2 gives probability 𝛄 of "staying at x" (or transitioning from x to x). However, this seems to be different than the characterization of lazy walks the paper gives, that of a fixed probability of 'halting' the walk. The effect of 𝛄 is rather to delay transitioning away from x for one time step. Is this discrepancy between halting and simply pausing important to the intent of the algorithm?	Gregory Todd Williams
Debbie Rank Typed Graph Walks Presentation Gregory Todd Williams 2	The model presented in the paper assumes every node x has a type τ(x). This model is clearly different from the RDF model in which a node many have zero or more types. Can the ranking approach in the paper be modified to support more flexible graph models such as RDF, or is the assumption of unique node typing fundamental?	Gregory Todd Williams
Debbie Rank Typed Graph Walks Presentation Joshua Shinavier 1		Joshua Shinavier
Rank Typed Graph Walks Ankesh	In the paper under local approach weights are assigned to edge types. i.e. if nodes x and y are connected by edge label l irrespective of x and y this edge has same weight for all such x and y. My question is, that by this aren't we loosing some local information? Saying two FBI agents are friends has very different significance from two terrorists being friends and much more when a terrorist and FBI agent are friends (it may not be realistic, is used to only convey the idea). Could loosing local information, mentioned above, be one of the reasons why local approach has not performed as well as global? For the given evaluation isn't it easy to come up with useful 'edge-label n-grams' or 'top edge-label n-grams' that could significantly boost precision after learning? i.e. Do you think that, since the relations that evaluation is looking for in the individual datasets are very few and narrow, it is easy to come up with few n-grams that can help improve the precision?	Ankesh Khandelwal

Absentees

Tim Lebo