Topicalizer Blog

Under http://www.topicalizer.com/augmentKeywords/ you can enter a sentence, phrase or word (you could possibly also enter longer texts, but the generated results would most probably be too large to make sense), whose keywords will be extracted along with their synonym / hyponym / hypernym relations by using WordNet.

This new feature is also part of the Topicalizer API.

There is a new screencast about Topicalizer and how it can be used on a blog called 'Sawfish Mango' (http://www.sawfishmango.com/), please have a look at http://www.sawfishmango.com/screencasts/mashup_paper.html for the screencast and http://sawfishmango.com/?p=19 for the related article.

Topicalizer now detects and displays custom document character sets, where before it simply assumed everything to be ISO-8859-1 or UTF-8. If no custom character set is detected or if plain text is submitted, UTF-8 is used instead.
This improved behaviour regarding character sets leads to an improvement concerning automatic language recognition and the overall analysis as well, especially for languages that contain many special characters like for instance, Norse languages or Russian.

Topicalizer is now listed at http://www.buzzshout.com/ and http://www.neobinaries.com/:
http://www.buzzshout.com/listing.php?name=Topicalizer
http://www.neobinaries.com/apps/Category/Miscellaneous/Topicalizer.aspx

Buzzshout and NEO Binaries both are websites which present new products related to what is now commonly known as 'Web 2.0' and have them reviewed by their community.

There was a bug that prevented stopwords from being recognised correctly when occuring right before a sentence boundary. This issue has just been solved, so stopword recognition should be even more reliable.

Today I have fixed a bug that occasionally led to JavaScript and CSS not being filtered before analysing an HTML document. Now JavaScript and CSS elements are removed correctly before the actual analysis is conducted, so they do not falsify the results anymore.

There is a new tool under http://www.topicalizer.com/similarDocuments/, which tries to find documents with similar content for a URL or a text by using the Google API.

Enjoy!