Do not use results when PubMed has zero results

[paulalbert1]

Background

ReCiter translates juc2004 into the following query:

(Charles-rawlins J[au] OR Charles-rawlins J[au] OR Charles-rawlins J[au])

This returns zero results using the traditional search, so PubMed "helpfully" searches just for the first initial as you can see here.

See how we're importing 276 records with the first name, last initial search?

Here's an example of record that's being imported.

This is the XML.

<eSearchResult>
<Count>264</Count>
<RetMax>1</RetMax>
<RetStart>0</RetStart>
<QueryKey>1</QueryKey>
<WebEnv>NCID_1_144258394_130.14.18.34_9001_1535657066_726671994_0MetA0_S_MegaStore</WebEnv>
<IdList>
<Id>30153686</Id>
</IdList>
<TranslationSet />
<TranslationStack>
<TermSet>
<Term>J[au]</Term>
<Field>au</Field>
<Count>264</Count>
<Explode>N</Explode>
</TermSet>
<OP>GROUP</OP>
</TranslationStack>
<QueryTranslation>J[au]</QueryTranslation>
<ErrorList>
<PhraseNotFound>Charles-rawlins</PhraseNotFound>
</ErrorList>
</eSearchResult>

Requirements

Do not import the resulting records if the attribute QueryTranslation is 5 characters and only 5 characters, and the final 4 characters are [au].

Then, do not import any records.

[sarbajitdutta]

Lets discuss this approach. Is this even needed at this point? @paulalbert1

[paulalbert1]

It is. I'm pretty sure the use case I gave above is still relevant.

[paulalbert1]

Another example...

[paulalbert1]

Another example...

[paulalbert1]

Another example...

Another example...

[paulalbert1]

The alternative is to put names in quotes. This returns zero results, but we may get more false negatives when people are being creative with their names.