Short Summary

With osoi speed,

11.00-13.00
Research Community Mining with Topic Identification
Information Visualization, 2006

Objective: Application for constructing network community from citation information and keywords (extracted from title only).


Tasks: 1)construct research model: where authors as nodes and kws as edges
2)specify interest research area by users: (given kw or set of kws), then nodes without edges contain given kw are discarded
3)community labeling: using frequent occurred kws (from ARM) as a property of community. Not simply count the occurrences, since the relationships between kws will be lost. Then, select papers written by authors in community and extract frequent kw pairs.

Points: 1) keyword map view to see relations of kws
2) chart view to see the number of kw occurrences over the time
3) community view to see network of authors in community with adjustable preferences

= = = = = = = = = =

14.00-17.00
A Trend Discovery System for Dynamic Web Content Mining
XI Int'l Conf on Computing CIC-2002

Objectives: Compute distribution of topics on NEWs over the specific period and identify the topic changes for the two consecutive periods


Pros: Interesting idea on simple distribution calculation with domain independent statistical analysis but reasonable.
Interesting results of topic identification by using method from "Interpreting Nominal Compounds for IR, L.Gay and W.B. Croft, Information Processing and Management, pp21-38, 1990", use POS tags and join nouns to form unique item when appearing in the sequence, select most frequent items.

Cons: No further extension on how to deal with other kinds of period, evaluation, etc.