Explorative visualisation of usage log data

Speaker: Markus Kirchberg

Using explorative visualisation to extract information/knowledge (usage patterns or relevance indications) from data and relate to users/real-world events

Logged data are generated by almost all systems/services, ability to extract value from logs is a key distinguishing factor.

Sample events: conference, floods in Singapore

Types of usage logs: network, firewall, vehical usage, database usage, web, ftp, mail, …

This work focus on http logs.

How to obtain meaningful web usage data: usage log analysis, social tagging, explicit user feedback

Usage log analysis is non-invasive and implicit.

Two data sets: Semantic Web Dog Food and DBpedia

First remove logs generated by scripts/bots (periodic, high rate).

Anonymise logs by removing IP address but add in country of the IP and hash of the IP.

Look at fan, depth and weight of Web Travel Footprint of each IP over a specific time window. Draw a Kandinsky graph to visualise the WTFs.