Twitter is rolling out annotations where you can associate meta data with your tweets. Do you people think this is going to help semantic web in a big way? What do you all think are its implications?

asked Jul 19 '10 at 04:21

ArchieIndian's gravatar image


3 Answers:

Annotations could be a big win for twitter. It will allow tweet data to be aggregated and indexed much more easily, and it will encourage users to review things and sell things via twitter. Both of these (though especially the latter) lead to significant monetization opportunities for the platform.

Personally, I have exceedingly low expectations of the semantic web -- 1) content creators don't want to write meta data, 2) content creators won't conform to meta data standards and 3) spam spam spam.

I think it's much more likely that some small set of data will be annotated with something relatively ad hoc (maybe like twitter annotations). The mapping from content to annotations can be learned (say, 140 character tweets to ratings) and then applied to unannotated data. From that perspective, these annotations could be very useful in sentiment analysis, IE, topic clustering, summarization, etc. of twitter posts. But I wouldn't go so far as to pin semantic web hopes on it.

answered Jul 19 '10 at 09:20

Andrew%20Rosenberg's gravatar image

Andrew Rosenberg

I see a lot of recent work in providing annotations for the web. EmotionML is a step towards including sentiment/opinions for documents on the Web. Sentiment Analysis and Opinion Mining has garnered huge interest among the research community in the past decade and there has been an exponential increase in research trying to build systems/applications to automatically understand and detect sentiments. It looks like there is a thrust to include sentiments/opinions as "annotations" for documents/posts (like tweets) to achieve the Semantic Web goal especially considering EmotionML is a working draft on W3C.

answered Sep 23 '10 at 05:31

Dexter's gravatar image


I think it's a good move and it would enrich the web content with lots of contextual knowledge which will lead to better usage of multi-view learning algorithms, similar to the prior work on harnessing social bookmarking data, and in general user-generated content, for mining web data (e.g., using social tags to improve web page clustering: see this paper and this paper).

answered Sep 23 '10 at 10:19

spinxl39's gravatar image


edited Sep 23 '10 at 10:21

Your answer
toggle preview

powered by OSQA

User submitted content is under Creative Commons: Attribution - Share Alike; Other things copyright (C) 2010, MetaOptimize LLC.