RPGExplorer began as a project for tinkering with natural language processing. Specifically, I used RPG blog articles as test case for extracting informative keywords in an attempt to categorize them logically. The categorization aimed to select non-site specific tags for articles for grouping related articles.
To date, the tagging effort has produced mixed results. In some cases, it does well. Other results are almost non-sensical. Providing interesting and accurate category results will require more programing and research. NLP is not my forte -- just something I am interested in.
The actual website was generated to visually look at the results. As of this writing, roughly 350 table top role playing blogs are monitored via RSS/ATOM feeds. The underlying database has accumulated 6500 posts. Only two weeks of posts are presented on the site to keep the data usage smaller. The full dataset is being retained offline for research and testing.
The crawler obeys the web robot protocol so you can restrict it from crawling anything on your site via modification of your site’s robots.txt to exclude ‘RPGExplorer’.