News

PSS! Build Your Own Topical Search Engine!

A company called Septet Systems has recently announced the alpha of a new service called PSS, short for Personal Search Syndication. PSS allows you to set up a personal search engine based on queries you specify. While I like the concept, there were a few things about it that made me feel a bit stifled. Right now it’s first-come first-serve for accounts; you can sign up at http://www.pssdir.com.

Sign up and agree to a Terms of Service but I can’t find a privacy policy anywhere… grrrrr… Once you’re signed up, you’ll be stepped through the process of building your search engine. I made one while I was signing up, but then I thought about all the folks I met at the APRA conference and decided to make another one, which I’ll walk through here.

The first step is to name and describe your PSS, give it a URL, and decide whether you want to make it public or private. I named mine Donors and Donor Rolls, with the thought that I’d try to build a search engine that indexed gift rolls and donation lists. The next step is to select a category and subcategory; unfortunately there’s nothing that really fits. I started with News/Education and settled on People/Arts.

After you select those two you’re given a series of queries boxes to fill out. The idea is that you fill out the query boxes, and PSS goes and creates an index of pages based on those queries.

The snag is that you have NOTHING in the way of flexibility. You can’t search in titles, for example. There doesn’t seem to be a way to restrict searches to a top-level domain. There’s no NEAR operator. I know I do a lot of advanced searching but being restricted to JUST search terms made me feel hamstrung.

But I’ll continue because I’ve already spent 20 minutes playing with this site so why not. I put in five very basic queries: “donor roll”, “gift roll”, “list of donors”, “thank our donors” and “our donors” alumni. (I’m sure actual prospect researchers could come up with a much better initial query list.) You’re also allowed to exclude query words from your searches; I excluded “venture capital”, IPO, and blood (because I didn’t want to get a list of blood donation sites or thank-you lists.)

You have the additional option of specifying categories, which provides a way for PSS to organize your content, but I couldn’t think of a good way to do that in this case. So I clicked Continue and I was finished.

Unfortunately once you’re finished you can’t jump right into looking at your PSS; the system requires about 30 minutes to scrape and build your search engine. To see how the attempt at a search engine for donor lists turned out, check out http://www.pssdir.com/dir/donors.

(Julia Child) Earlier today I baked a PSS at 350 degrees for 30 minutes. (/Julia Child) The one I baked was called “Hypothyroid” and was an attempt to gather information centered around hypothyroid disease. You can visit it at http://www.pssdir.com/dir/hypothyroid.

The search engines have a limit of 1000 documents, so you’re not going to get absolutely everything out there. The list of pages in your search engine includes title, size, date indexed, summary, a link to view the cache, and two small icons to indicate the type of document it is (the HTML icon is a little IE icon, which was very confusing for a moment) and whether or not anyone has commented on that page. For the most part I was pleased with the relevance of the indexed content, until I came across “GameDAILY Blog – How To Not Armor Your Horse.” PSS had indexed it based on comment spam. Fortunately there’s an easy way to get rid of pages; click the checkbox next to them and then click on “Delete selected links”. Horse armor commentary all gone.

Obviously you do not get to delete links out of a PSS unless you’re the one who built it. But if you are the one who built it you have other options, too, like excluding specific documents and entire domains out of your index. Kind of annoying to have to do this after the fact; can’t the domain exclusion at least be integrated into the PSS building process?

Once you’ve built a search engine you like and removed all the irrelevant barding badinage, you can put your PSS to work. You can create “search alerts” that will send you a weekly or daily e-mail when the query word you specify is found in an indexed document (Click “Search Alerts” at the top of a PSS’ page.) You can also put a search form for the PSS on your own site; click on the oddly-named “Search Stub” (how about “Search Form”?) at the top of a PSS’ page. You’ll be taken to a page with a very, very basic form for searching your PSS. Here’s what it looks like.

Search Donors and Donations:

With so many solutions out there that allow you to specify a list of domains to search, it’s good to find one that takes a different approach and starts you with a list of queries. However, if you’re any kind of power searcher, the inability to use special syntax will get real frustrating real quick. I would also integrate some of the post-building configuration, like excluding domains, into the building process. If the beta adds special syntax and expands indexes beyond 1000 pages, I will be a real happy PSS user.

Categories: News