Sentisquare software gathers comments related to a brand from
various sources. The comments are clustered into
topics using semantic
analysis, which is able to link two opinions even if they are
expressed in different words. Thanks to this, Sentisquare can
compute how many contributors wrote about a topic and prioritize
For each topic, it automatically produces a summary of key opinions by
selecting the most important text spans. The vector
representation enables to link
topics across different brands. Thus we can see how unique the
topics are for a particular brand. It also provides temporal linking of topics. We can
see then how topics develop over time.
On the top of the above, we use unique method that deals with complex morphology of a language.
Thanks to this, Sentisquare is able to work even with less
frequently used languages (such as Czech language).