The Memento Web and the Wayback Machine are two possible solutions:
The Wayback Machine by the Internet Archive is your best friend for all things that were once online, and even some things that still are, if you want to compare changes.
The Wayback Machine is a digital archive of the World Wide Web and other ...
The first source of raw interactions that fit your needs that comes to mind is Twitter.
NCSU's Tweet Sentiment Visualization
http://www.csc.ncsu.edu/faculty/healey/tweet_viz/tweet_app/ seems pretty good in that it allows you to enter keywords and get a graph of recent tweets graphed on axes (pleasant/unpleasant and active/inactive). It also has some pretty ...
You may want to stop and ask why you should want to be indexed there.
Dataset Search seems like a poorly planned effort to try and corner dataset search. Why the push isn't for open linked datasets/integration with search reveals many details. Performing a search doesn't even tell you how many results there are.
Reddit AMA - Chris ...
Although not strictly open data, the in my experience most fruitful way to search for something like this is:
Search for the raw data (timeseries, tables, values) in studies/statistics oneself and prepare the graph oneself or
do a simple Google image search like graph hdd storage price.
I suspect this is not supported. There's no evidence of it in the "advanced search", nor in the guide to search operators (which advises against worrying about memorizing all of the operators because you can use the advanced search page.)
Also, the advanced search help page lists filters and doesn't include file size.
I have actually been working on something very similar; a search engine for restaurants in London. The search engine crawls popular social media platforms (TripAdvisor, Open Table etc.) on a daily basis and allows users to get an overall view for a given restaurant based on written comments from EVERY review.
You can also search for a particular dish / ...
Meta-data is more important than ever.
Google and other search engines created their own meta-data and vocabularies over at http://Schema.org.
Google would for example not be able to show you runs of movies in the next cinema in the way it does, if the website didn't offer that as machine readable data (see MovieTheater and Movie and related concepts).
The documentation of the Twitter Search API for tweets mentions operators, such as ":)", where you can specify whether the tweet has a positive or a negative attitude. You can combine this with the operator that asks for tweets that contains questions ("?"), and with any other operator such as hashtag search, e.g. #London, or #song
While I don't work for The Post or Google, and thus cannot speak authoritatively, I was at a Google event for top news organizations last year in NYC, and can attest to the fact that -- depending on the publication -- Google often provides custom Trends data as an enterprise service to publishers. That would be my guess here.
If you're interested in Twitter search, you can prep your own tweet IR collection using the open tools described on the TREC Microblog (MB) 2011/2012 pages. The test queries and judgements are also openly available from TREC, as is the tool for evaluating (trec_eval)