Semantic Web tech. I know, I know... it's easy to sit back and say "WTF, the Semantic Web is a pipe-dream, and it's been being developed for 15 years and it isn't here yet" etc., etc.<p>I will counter that by contending that the Semantic Web <i>is</i> here, just not fully so, and furthermore, that a more and more complete Semantic Web is inevitable.<p>My feeling is that you couldn't go wrong learning RDF, SPARQL, etc, AND, more specifically, focus on NLP. Why NLP? Because NLP is the bridge between the gazillions of petabytes of "stuff" that's out there that <i>isn't</i> part of the Semantic Web, and the structured data that does make up the Semantic Web. When you can use NLP (and related tech) to extract semantic meaning from free-form text and then make it part of the Semantic Web, that's pretty powerful stuff.<p>Take a look, for example, at Apache Stanbol[1] and the stuff they're doing with extracting structure from text.<p>Also, look at things like dbpedia[2] and the Linked Data[3] initiative. Seriously, seriously cool stuff is going on...<p>[1]: <a href="http://stanbol.apache.org" rel="nofollow">http://stanbol.apache.org</a><p>[2]: <a href="http://www.dbpedia.org" rel="nofollow">http://www.dbpedia.org</a><p>[3]: <a href="http://linkeddata.org" rel="nofollow">http://linkeddata.org</a>