Both stemming and lemmatization allow queries to match different forms of words. Stemming was commonly implemented with Reduction techniques, though this is not universal. Lemmatization implies a possibly broader scope of functionality, which may include synonyms, though most engines support thesaurus-aided searches in one form or another. Lemmatization also tends to be expansion based (either index or query time), though this is not universal. Word variation rules that rarely change are often implemented with reduction or index time expansion, while rules that periodically change are best served with query time expansion.
We've been enjoying our new search engine, FAST ESP 5.3, and it's certainly a very powerful platform. Today on our development server we can't seem to spider any more pages, and I'm seeing a weird error: "session_servant (27): DocAPI manually suspended. Operation denied."
We are installing FAST ESP 5.2 on Red Hat GNU/Linux Kernel version 2.6.9-55.ELsmp. But the install fails as per the logs below. The java version is "1.5.0_15" and it seems to be supported, but we see errors in the logs.
Can you help us out with some clue as to what could be wrong with our installation process?
One of the exciting features of FAST ESP 5.0 is the new interface intended for the business user, the Search Business Center. This new UI allows business users to view reports and make changes to layout and vocabulary items without needing to pester the IT department. The reporting subsystem is very useful. It shows search activity in both report and graph form, including popular searches and searches that returned no results.
But Dr. Search had a problem... it wasn't working! The good doctor even tried upgrading to the latest version, ESP 5.0.7, and still no luck.
This month one of Dr. Search's associates was in a hurry to install a copy of FAST ESP 5.06 for a demo using Windows Drive D:. During the install, the program suggested that the data directory might best be located on C:, the implication being that the program and data directories should be on different spindles. The associate, who knew the system drive didn't have as much room for data and collections as D:, decided to re-start the install, placing the program files on C: and the data files on D:. As it turns out, that was a mistake.
We've said for years that the secret to great search relevance is great indexing. The problem is, with many enterprise search engines, it takes a good deal of effort to intercept and improve the document indexing process. Sometimes you want to add extra metadata from an external source, or you want to clean up the field values that will be populated automatically during indexing. Some technologies, like Hummingbird's Fulcrum Technologies, allowed developers to write C code to sit in the indexing process; and Ultraseek has long had the patches.py code to add custom code to pre-process each document in the indexing process. But generally it's been a challenge.