###link:
A live demo can be found here.
A video demo can be found here.
The final paper of this project can be found here.
###Project Summary:
This project aims to produce an easy and fast way for users to create taxonomies through interaction and visualization.
Nowadays many people uses the social media to express their opinion or discuss about many topics. This data could reveal the way how people around this world react to social issues such as environment, pollution, and sea level rising problems. One problem is that it is hard to filter which documents are interesting and which are not. For example, which words belongs to the topic “climate change”. An easy way would be searching for texts that mention the term “climate change”, the problem is that some documents may talk about climate change but do not mention the word, for example they could mention “warming planet” for example.
To solve this problem many approaches have been proposed many of them relying in machine learning algorithms. But the problem with machine learning approaches is that they (1) needs specialized data mining staff (2) do not provide a clear description of what belongs to the topic and what not, not allowing the domain expert to easy interact with. Another option is to create a taxonomy for the topic, that is, a description of which words belong to a topic and which ones do not. This taxonomy is then used to perform a query in the documents and get only relevant documents.
###Project group member:
Cristian, Peixin, Wei