? PubSub unveils new government search |
| Woodward and Bernstein, bloggers? ?
June 01, 2005
Help: Tips to set up PubSub feeds
I felt like a jerk through the entire morning commute. I wrote in a breakfast post that PubSub feeds could be tricky to engineer, and I mentioned in parentheses that I had relied on help from the company. The implicit message: I get help from the company; you can fend for yourselves. Sorry about that.
We're all going to have to learn the ins and outs of this type of prospective search, just the way we've learned Google over the past years. For starters, here are PubSub's tips for sculpting your own feed searches--and sifting out the junk. But if any of you have further tips or ideas on the subject, or advice on other prospective search engines, please send them along.
TrackBack URL for this entry:
I remember the days when the only way to use a retrospective search system was to find a librarian who would interview you to determine your needs and then do the search for you. Then, in the 90's, sites like Altavista and later Google broke through the usability issues and made the interfaces massively easier than they had been before. Still, it took some time before people built up enough experience to be comfortable with routinely doing something that had previously been a specialist technical task.
We're trying hard to make prospective search just as easy (or almost as easy...) as retrospective search has become, but we've clearly got more learning to do. So far, I must admit that our primary focus has been on making the system work. We've only recently had the luxury of working on the usability issues rather than the very difficult technical problems of building a system that can do matches at the rate of billions per second... PubSub is the first Internet Scale prospective search engine and thus we don't benefit from having others' experience to copy. But, we're working hard on a number of ideas to improve the usability of the system. My hope is that in future you'll be able to write a post saying something like: "Once prospective search was hard, but we've learned how to make it easy."!
If you have any ideas on how to make it easier to use, please send feedback to firstname.lastname@example.org.
Posted by: Bob Wyman at June 1, 2005 11:52 AM
Don't forget the misspellings. When setting up a PubSub feed for a name or word that has common misspellings be sure to include them with an OR statement. For example the ego-feed I watch is Hallett OR Hallet. When Buzz Bruggeman was doing his ego-results-research on Google vs. MSN I told him to also search for "Buzz Burggerman" turns out there were over 200 results with the misspelling of his name.
Posted by: Josh Hallett at June 1, 2005 12:34 PM
My experience has been pretty good thus far. There was a bit of a learning curve figuring out how to refine the keywords I wanted to monitor without getting a bunch of non-sense also.
The one thing that helped me weed out the majority of the non-sense was to use the exact phase search (put your search terms in “quotes”). For my purposes a search term looks like this “Internet Security” as apposed to my original term security that returned a bunch of military security type information. While that is good it wasn’t what I specifically was looking for.
Posted by: Travis Metcalf at June 1, 2005 12:43 PM
Travis' suggestion that you use quotes to identify phrases like "Internet Security" is *very* important and will really help you to improve the results returned by PubSub. This quoting syntax works, of course, on the retrospective search engines as well but research indicates that even such a simple mechanism as this is more "complex" than the vast majority of search engine users are comfortable with.
One thing that those of us "in the business" need to remember is that even those things which seem simple or even trivial to we who work with the stuff every day can seem complicated and obscure to folk who do something other than computer science as a business!
For the computer geeks, don't forget that paranthesis are useful on PubSub. Thus, you can say things like A OR (B AND C)... How do we make *that* simple to use?
Posted by: Bob Wyman at June 1, 2005 02:16 PM
Bob, if you have the time, I think it would be helpful if you or someone at PubSub showed us how you built the search for, say, Sandra Day O'Connor. You must have included devices that you're not interested in the "day" of the week, and you don't want Sandra Bullock, and it should have something to do with law, etc. etc. I think if we see an example like that, we'll have more luck with our own variations. thanks, Steve
Posted by: steve baker at June 1, 2005 02:30 PM
The search for Sandra Day O'Connor is actually rather easy:
"Sandra Day O'Connor"
with the quotes will only return blogs with that exact phrase, if you wanted to broaden things a bit you could use
"Sandra Day O'Connor" OR "Sandra O'Connor"
Sandra Day O'Connor
will, as you predicted will return blogs posts with and 'Sandra' and any instance of the word 'day'
Posted by: Josh Hallett at June 1, 2005 04:37 PM
I am a librarian at PubSub and the one that created the queries for the government feeds. The query for Sondra Day O'Connor is "Sondra /within 2 O'Connor".
We have recently started to support the /within syntax. At this time, it can only be used for two separate keywords and not phrases. Thus, we don't think that the /within syntax is easy enough for non-technical people to utilize, due to its restrictions.
For the O'Connor feed, this query will pick up on any posting with the word Sondra within 2 words of the word O'Connor, thus eliminating the need to use Josh's query (although both structures will produce the same results).
As an aside, if you want to see any query that was created for any of the government subscriptions, click on the green plus sign next to each one, add it to your account, and then edit it.
I hope this helps.
Posted by: Steven M. Cohen at June 2, 2005 09:45 AM
I just realized that I spelled Sandra incorrectly in my previous comment. Yet another lesson: If the name/company/whatever is sometimes misspelled, make sure it goes into a query.
I've since updated the Day O'Connor sub to:
"Sondra /within 2 O'Connor" OR "Sandra /within 2 O'Connor"
Posted by: Steven M. Cohen at June 2, 2005 12:14 PM
Good information from the source - Thanks!
I was unaware of PubSub's existence until I talked to an analyst from Jupiter Research and he told me I should be monitoring keywords for my blog.
It has become a useful tool.
Posted by: Travis Metcalf at June 2, 2005 12:29 PM
One thing I'm doing for a client is setting up RSS feeds of persistent searches in multiple engines and running them all into one folder in a feed reader. Pub Sub is cool, but why rely on it alone when searching Technorati, Yahoo News, Google News, GoogleAlerts and lots of other sources can be done at the same time? I know I get unique results between all of them.
Posted by: Marshall at June 4, 2005 06:02 PM