Can anyone reassure me?
Or am I in an old-fashioned mind-set believing as I do that search engines
should consistently retrieve ALL articles in a dataset that match the search
criteria? And should give the same answer if the same search is repeated
(irrespective of whether the search is "manual" or "automatic") even if it
is repeated with a different search engine.
Or are we in a new fuzzy age where records are statistically matched and on
the balance of probability will approximate to what the system determines
that the user wants? [my most charitable explanation for observed anomalies]
I have documented discrepancies for Dialog on many occasions over more than
a year between results obtained on Profound retrieved by an alert and
results retrieved running the *same* search in manual or test mode. I have
even referred it to CIQM. But I have received no explanation to my
satisfaction. [I *know* that an alert can't retrieve a document BEFORE it
has been added to the database - but it should retrieve ALL relevant
documents AFTER they have been added]
It can't just be me, can it???
Could some kind soul with Profound Alerts already set up PLEASE help me out
by doing a simple experiment:
1. In Alert Manager expand one of your profiles and print out the title list
[Assumption: this will include all recent retrieved articles sent by
e-mail. An assumption I believe to be valid after comparing on many
occasions with titles actually received]
2. Running the profile in test mode [still in Alert Manager] and printing
the new title list.
3. Disregard the answers retrieved by both profiles within (say) the last
week, and compare the 2 title lists.
Are there any titles found in the "test" list, not found in the alert list?
Are they relevant (i.e. should they have been found by the Alert?)
TIA
Dave Huddart email: [log in to unmask]
Information Coordinator tel: 0121 541 3304
Oldbury Library fax: 0121 541 3346
Albright & Wilson
Trinity St
Oldbury, W Midlands, B69 4LN
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|