Let AI Learn from Web 2.0. Tag Co-Occurrence based Text Categorization as an Example

T. Lints (Estonia)


Artificial Intelligence, Knowledge Acquisition, Text Cate gorization.


The paper draws attention to the potentially high value of user-generated web content from the viewpoint of Artifi cial Intelligence. Web 2.0 sites can be used as a quickly growing base of knowledge to learn from automatically. To motivate researchers to make use of that data source, a demonstration is provided about the ease of obtaining and using tag co-occurrence data from a social bookmarking site. The described demo uses this derived info to catego rize texts into predefined categories and, according to sub jective evaluation, does so sensibly well. Though, being just a simple demo case for raising AI researchers’ inter est in Web 2.0 data usage, it is definitely not meant to be comparable to industry grade text categorizers in its current state.

Important Links:

Go Back