The Technology of Search
- Anatomy of a Search Engine
- Clustering Results
- Databases vs. Search Engines
- Search 2.0: Beyond Single-Shot Relevancy
- Search Engine Security
- Virtualization and Search
Technology is talked about quite a bit in the search space, complete with bragging rights about patents PHDs, and lots of overused buzzwords. Sometimes it's tough translating between "vendor speak" and client business needs. To be fair, there have been many great advances, there are many new features on the market, and at more attractive prices. But whether you should opt for unsupervised clustering or dilithium crystals depends in large part on what your company is trying to accomplish.
He's some of the initial inquiries we get:
- Should we just buy a Google box?
- Our search sucks, which vendor should we buy now?
- Do we need a taxonomy?
- Is "search engine performance" the same as SEO?
- What about that "open source stuff", should we just get that?
- We want a search engine with a shopping cart module - certainly a reasonable request.
- Look, we wrote a 50 page RFP!
Meanwhile, this is how many of our chats with vendors commence:
- We have autonomous goal seeking uber clusters, with mnemonic N-tuples!!!
- We have so many patents! / We have so many PhDs!
- We're going to be the "Google of China" !
- We're gonna build a partner network ecosystem for our engine.
- Nobody else can do this!
Most of these sound reasonable, especially from the client side of things, but you'll notice there's very little overlap between these two lists.
Keep in mind that technology alone only goes so far, we like to say that "there are no magic beans" - what we mean is that nothing can make up for missing data, horrible UI design, or a completely untrained staff.
Key technical issues:
- Overhyped features, though there are some good ones out there.
- Buzzwords and abbreviations, but what features do you actually need?
- Unrealistic expectations - even humans don't agree on the "best" answer to some questions
- Conversely, apathy towards new technology that does work
- Trying to replace human intelligence with machines, vs. augmenting and amplifying it.
- Trying to entirely replace business analysis and process with technology
- Search engine mismatch - the wrong engine for the job, a painful discovery
- Staffing mismatch - different products require different staffing
- Vocabulary mismatch - fortunately this can be fixed!
- And finally, packaging, Packaging, PACKAGING! - it needs to install and spider data