Lexalytics Leverages the Human Element of Wikipedia

April 26, 2011

Whether you embrace or scorn Wikipedia as a reference, there is at least one good use for it, according to Lexalytics.  “Lexalytics Analyzes Wikipedia to Understand How Humans Think” describes the rewards of their idea for an alternative function for the user-edited online dictionary.

Lexalytics has turned its ability to process and summarize large quantities of data on Wikipedia in an effort to improve sentiment analysis.  They “used it as a source for how people think about the organization of information and for perspective on how bits of information are related to each other.”  The ultimate goal: the creation of a “concept matrix” to bolster a machine’s ability to link like ideas independently, an essential task in human language comprehension.

The results of the project showed the machine could connect that both a rose and a daisy were classified as flowers.  If that isn’t baby steps, I don’t know what is.  Granted, how does the saying go… you have to walk before you can run?  Machine processing of human language is a difficult undertaking; and why shouldn’t it be considering human processing of human language at times can be an equally complicated and oft-failed task.

The article mentions most of the confusion still lies in sarcasm and double entendres, some of my personal favorite aspects of language.  So looks like I may have to wait for the next version of the software. Wikipedia seems to be emerging as the corpus of choice to demonstrate search and content processing technology.

Sarah Rogers, April 26, 2011



Comments are closed.

  • Archives

  • Recent Posts

  • Meta