Memoplex Browser Searching and Browsing in Semantic Networks

Memoplex Browser: Searching and Browsing in Semantic Networks CPSC 533 C - Project Update Yoel Lanir

Motivation n Semantic (associative) networks is a good way to store information When retrieving information from large data sets (search) the user can use browsing as an addition to search Browsing through associative knowledge is how our brain works

Goal n To Assist the user to search and browse a large dataset using a browsing tool which will navigate efficiently through the semantic ties of the dataset.

Domain n 2000 articles extracted from the New. York Times. Each article contains: n n Title Author Publication date text

Proposed solution n Two levels of viewing: n n Node level: show a radial view, which puts a single node in the center. Similar to Yee et al. (2001) with addition of weights on edges. Show only two or three levels from the focused node building the tree according to a breadth first algorithm. Keyword/category level

Node level n n n Browsing history window (linked to the node view) Details window Labeling: Excentric labeling (Fekete and Plaisant) Animation Scalable Search window

Keyword (Category) level n n Each document has a list of keywords with weights which is used to classify the document. Keywords can be looked at as categories! Will show a graph of keywords linked according to number of articles sharing categories. (2 categories with many articles belonging to both will have a strong link) Semantic zooming between category and node level

Project state ü ü ü n Learn Prefuse tool Prototype of node visualization Discuss design requirements Finish design of keyword network n n n Connection between keywords How to represent the keyword network How to map between nodes and keyword network in both directions Implement windows and links in node view and connect to real data Implement keyword network and semantic zooming