Quote-Back Technology
(Previously called Sub Page Reference)
Contents
Why are we doing this
We are a site about sites (among other things.) As such, we desire to cite web pages with a specificity greater than provided by a url alone.
- We cite where we find paragraphs that we include on our domain pages. This information is now just text, not information that is systematic in any particular way.
- We extract numbers from social software sites for inclusion in our Presidental Portal. We have automated this collection of over 100 numbers but have not done so in a way that is particularly robust or subject to community adjustment.
- We anticipate a refinement on the web service tinyurl.com that we could build on reliable sub-page references. We would like to use such a service as an entry-level service that could be useful for bloggers or anyone else that writes about the web.
We'll know when we are done
There are many ways such a service could be implemented. We would like to explore most of them over time. However, for the purpose of this project we will be happy if we achieve the following:
- We produce an external link that can be pasted into one of our domain pages, that, when clicked, takes one to a version of the source page that has the specific portion of the page highlighted.
- We record enough supplemental information about the source page in our own databases that we can find the original location even when the source is updated, or, should that prove impossible, we can give a reader useful information for their own search.
- We provide some workable ui, possibly as a browser plug-in, for the construction of the external links mentioned above.
Use Cases
These are use cases that can exploit this technology.
(This zip file contains screen shots used to mock up each of these use cases.)
Collecting Thoughts
wikitect is a small group organizing for a purpose. The founder is blogging about it and many other things on his blog. We'd like to circle the blog phrase that begins My proposal is for the development of a collaborative tool that supports, in a structured way, the development of pattern languages ... and copy that into the What We Do section of the wikitect page.
A variation on this would be to read a long blog post and collect several quotes to be brought back to AboutUs in one operation.
Extracting Numbers
The Portal:2008_Presidential_Election rolls up stats collected from numerous dynamic web sites. For example, Barack Obama's Eventfull page reports that 28582 people would like to hear him speak. When the AboutUs page goes stale, we'd like to update it automatically, perhaps by selecting the region with the numbers and saying to get the first proper number from that region.
When Eventfull switches to a new page format, we'd like to gracefully fail and solicit community help in finding the new number location. It could be on a different page, or in a different location on the same page. (We'd call this tending the scraper.)
A variation of this could take a parameter, say a candidates name, and scrape that particular page for the familiar content. This is the way the current Presidential Scraper works. This would be doing more than just quoting, it would be creating an ad-hoc web service.
Sharing Shorthand
A popular site such as [BoingBoing] blogs about another thoughtful writer such as Kevin Kelly. We offer bloggers a way to reference the quotes they take and include them on their own blogs, not just on AboutUs pages. We allow some shared branding of our widgets in this case, perhaps just changing color or maybe adding extra buttons of their design.
A variation of this would be to create permanent short links in the style of TinyURL.com that would allow similar references to content in any medium such as email and even read over phone calls.
Related Work
- MIT's Simile project has a web scraper called Solvent that is a Firefox plugin.
- Google Notebooks (faq) are similar with a capture tool and sharing capability.
- http://www.netsnippets.com/ is mentioned in the Google Notebooks forums.
- the netsnippets site mentions http://www.esnips.com/ (but doesn't quote content, only holds it)
- Dapper seems to compare samples of desired content in order to analyze the page structure. See http://www.dapper.net/dapperDemo/
