Inconsistent Results with SOAP API
dkerins
contains a few documents with fairly unique strings. When I do a
regular google search with the following term...
site:pce.bitbox.ca A0ACA011 B76687DE
I get the expected results. (i.e. There are 4 pages with these two
terms not including the index file.
When I do the same query through the SOAP API my results are not
correct.
i.e. I get two of the 4 files and the index file. I get this
incorrect result using a perl script and with the java that came with
the API too.
Should I not expect consistency?
If you need more info about what I am doing to get the inconsistency,
let me know.
Thanks for any insight to this perplexing problem..
David
Manfred
> I am doing some research work for a thesis. I have a site which
> contains a few documents with fairly unique strings. When I do a
> regular google search with the following term...
> site:pce.bitbox.ca A0ACA011 B76687DE
>
> I get the expected results. (i.e. There are 4 pages with these two
> terms not including the index file.
>
> When I do the same query through the SOAP API my results are not
> correct.
> i.e. I get two of the 4 files and the index file. I get this
> incorrect result using a perl script and with the java that came with
> the API too.
>
> Should I not expect consistency?
interface _and_ the api.
For example, when I use your query in the web
site:pce.bitbox.ca A0ACA011 B76687DE
I get (apart from the index) 3:
pce.bitbox.ca/a-and-b-and-d.html
pce.bitbox.ca/a-and-b.html
pce.bitbox.ca/a-and-b-and-e.html
It depends on which Google datacenters are involved for the response.
Hope this helps,
Manfred
dkerins
Thanks Manfred.
Is it reasonable to expect that over time the various data centers
would synchronize and that their differing results would converge into
one, dare I say accurate, world view?
Thanks David
Manfred
Hi David,
With the query "site:pce.bitbox.ca" I get 13 pages, from which 8 pages
are already in the supplemental index. This means that convergence is
not likely to occur.
I would also question the assumptions (or model) about the search
engine which build the basis for your experiments.
Hope this helps,
Manfred
dkerins
Thanks for the insight. What is the supplemental index? Why wouldn't
the different data centers eventually have same info about a site that
is fairly static? Is there somewhere I could get more information
about this so that my expectations are in line with what google would
provide.
On the pce.bitbox.ca site there are actually 15 pages including the
index file.
The work I am doing focuses on creating the notion of user preference
with search terms.
This allows the user to require some terms and prefer (to varying
degrees) but not require other terms. My software does the
preprocessing and sends an series of queries to simulate this
specified preference.
So a preference query would look like this...
site:pce.bitbox.ca A0ACA011 U1[B76687DE] U2[C9B136AF]
means A0ACA011 is required and the other two are preferred but not
required (U1 is a more preferred term than U2).
My software would parse this and send a set of queries and then gather
and present the results to the user.
The google queries would be...
site:pce.bitbox.ca A0ACA011 B76687DE C9B136AF
site:pce.bitbox.ca A0ACA011 B76687DE -C9B136AF
site:pce.bitbox.ca A0ACA011 -B76687DE C9B136AF
site:pce.bitbox.ca A0ACA011 -B76687DE -C9B136AF
The more preferred/successful query is the first one, the last is the
least preferred/successful result.
As you can see the user can kill 4 birds with one stone, so to speak.
The pce.bitbox.ca site is used simply to prove that my software works
correctly, so having google return a consistent set of results would
be helpful. (Not that I am expecting you to "fix it" for me :-). I am
simply presenting this information to you so you might understand what
I am up to and why I have the expectations that I do.
Thanks so much for your input.
David.
Manfred
with a few hundred pages and
mark some (10-20) according to your needs. You can hide your markers
with css.
Pages in the supplemental index are those, which normally are not
shown in the search engine response page (SERP) only _after_ all the
other pages. You can check beforehand to avoid choosing a site which
has pages in the supplemental index.
Pages are thrown in the supplemental index because of almost no
content, duplicate content, problems of the search engine to parse the
content ...
Hope this helps, Manfred
dkerins
Hi Manfred, I am not quite sure how I would "choose a living site" and
"mark some according to your needs". Would I need to contact the site
owner and get permission to bury my markers in their pages?
Or can I somehow mirror a site and mark pages without changing the
"living site"?
I am just not sure what you mean here.
Apologies for being dense.
David
Manfred
> Hi Manfred, I am not quite sure how I would "choose a living site" and
> "mark some according to your needs". Would I need to contact the site
> owner and get permission to bury my markers in their pages?
> Or can I somehow mirror a site and mark pages without changing the
> "living site"?
No way. With "mirrored" pages you not only violate the copyright, but
you also get penalized by Google for duplicate content.
Manfred
dkerins
Alright, thanks for being so helpful.
I guess I could also populate my pce.bitbox.ca site with content ( I
have alot to say about preferential constraint expressions and
infinitesimal logic. If I spread that around my 15 pages perhaps that
might get the site out of the supplemental index?
David